Skip to content

200 results for β€œEvo”

ListGrid
What you seeeverything
RecentOldestAddedA–Z
All types200Articles83Papers57Write-ups26Repos9Advisories7
Talks6Threads6Courses2Tools2Other1Videos1
β€œEvo”
decrypt.coarticleπŸ›‘οΈ Security

Crypto Docs Host GitBook Hit by AI Agent Security Breach, Urges Users to Rotate Keys

GitBook disclosed an attack it calls "sophisticated, highly targeted" and AI-agent-assisted. The attacker gained unauthorized access to infrastructure credentials at 7 p.m. CET on 2 November 2026. Potentially exposed data includes API tokens, integration secrets, Git Sync secrets and AI assistant questions. GitBook revoked personal API tokens and told customers to rotate credentials and re-authenticate MCP. PancakeSwap moved its docs off GitBook. GitBook has not said how the agents were used.

GitBookModel Context Protocol
PancakeSwapOpenAIAnthropic

Threat Intel, AI SecurityΒ· Published and added 3 Nov 2026

thedefiant.ioarticleπŸ›‘οΈ Security

GitBook Revokes All Personal API Tokens After Breach

GitBook revoked all personal API tokens after a breach of its documentation platform. The company says the attack used AI agents and may have exposed integration credentials and questions submitted by documentation visitors. The fetched page held only the headline and subhead, so the intrusion method, timeline and scope are not available.

GitBook
The Defiant

Threat Intel, AI SecurityΒ· Published and added 3 Nov 2026

en.cryptonomist.charticle🧠 AI

Revolut’s AI cost reduction hits 90% with Chinese GLM models

Revolut reports cutting AI costs by about 90% after moving about 2,000 engineers from OpenAI and Anthropic services to Z.ai's GLM 5.3, self-hosted on 56 NVIDIA B200 GPUs. Head of AI operations David Abitbol cites $0.30 per million tokens and time to first token falling from 14.26 to 1.68 seconds between April and August 2026. FP4 quantization and KV cache tuning drove the gains. The report gives no deployment detail beyond these figures.

GLMNVIDIA B200FP4 quantization+1
RevolutZ.aiGLM-5.3+4

inference, ai-infra, ai-policyΒ· Published and added 3 Nov 2026

en.cryptonomist.charticleπŸ›‘οΈ Security

Binance’s $6.34B hack began with fake interviews and an AI tool exploit

Binance's initial report on the October 29, 2026 compromise of Binance.com, a $6.34 billion loss. Attackers planted malware on an engineer's device through fake job interviews in September 2026, abused an internal AI tool to reach wallet engineers, then compromised the wallet transaction signing service. Malicious EIP-7702 authorizations stayed valid after withdrawals were halted and needed separate revocation. The FBI published TraderTraitor wallet addresses.

Ethereum
BinanceTraderTraitorLazarus Group+4

Blockchain, Threat IntelΒ· Published and added 3 Nov 2026

decrypt.coarticle🧠 AI

Proton Users Revolt After CEO Calls Security Caution 'Loser Thinking' Amid AI Push

Decrypt reports on backlash against Proton after 404 Media revealed that CEO Andy Yen called a request for pentesting time "loser thinking" while pushing engineers to use AI instead of hand-writing code. Proton cancelled planned AI-use re-interviews on 30 October 2026, and CTO Bart Butler's 1 November manifesto says engineers should direct LLMs and that most human code review may end. User cancellations are anecdotal, not a verified count.

LLMLumoClaude+1
Proton404 MediaMicrosoft+5

ai-policy, coding-agentsΒ· Published and added 2 Nov 2026

securityweek.comarticleπŸ›‘οΈ Security

Revolve Data Breach Impacts at Least 180,000 People

Revolve, an online fashion retailer, disclosed a September 11 breach of systems including its e-commerce platform. Attackers may have exfiltrated names, addresses, card numbers and expiration dates that customers saved in their accounts. Notices filed with the Iowa, Massachusetts, Texas and Vermont attorneys general show at least 179,259 people notified. Revolve offers no credit monitoring, and the attacker and method are undisclosed.

RevolveSEC

Threat IntelΒ· Published and added 2 Nov 2026

bleepingcomputer.comarticleπŸ›‘οΈ Security

FBI links $6.3 billion Binance crypto heist to North Korean hackers

The FBI attributed the roughly $6.3 billion Binance theft of about 29 October 2026 to North Korea's TraderTraitor (Lazarus Group, Jade Sleet, APT38). Binance traced the breach to a compromised DevOps engineer's laptop that gave access to a service container used to submit unauthorized withdrawals across ten blockchains. The FBI published 628 laundering addresses and linked the group to earlier thefts at Bybit, DMM Bitcoin, Bitget and Drift.

EthereumBitcoinSolana+3
TraderTraitorLazarus GroupDPRK+14

Threat Intel, BlockchainΒ· Published and added 1 Nov 2026

decrypt.coarticleπŸ›‘οΈ Security

Binance Hack: Everything You Need to Know About the Biggest Crypto Heist in History

Decrypt's explainer on the roughly $6.4 billion theft from Binance hot wallets across about 10 blockchains. Binance traced it to a compromised DevOps engineer laptop plus further breach points, and on Ethereum the attacker used 50 pre-signed EIP-7702 authorizations to turn a hot wallet into a sweeper. Chainalysis and TRM Labs attribute it to TraderTraitor; Tether and Circle froze about $2.3 billion.

EthereumBitcoinTether+3
TraderTraitorBinanceChainalysis+10

Blockchain, Threat IntelΒ· Published and added 31 Oct 2026

stepsecurity.ioarticleπŸ›‘οΈ Security

Binance’s $6.3 Billion Hack Puts Developer Machine Security in Focus

StepSecurity's analysis of Binance's 29 October 2026 incident, which Binance puts at about $6.3 billion in affected assets. It covers Binance's disclosure of a possible DevOps laptop compromise, and TRM Labs and Elliptic findings on 50 EIP-7702 authorizations and a persistent delegation. It links this to the Nx, OVHcloud CLI and Claude Code extension compromises, gives an investigation checklist, then describes StepSecurity products.

EthereumnpmVS Code+3
BinanceTRM LabsElliptic+5

Blockchain, Supply ChainΒ· Published and added 30 Oct 2026

theblock.coarticleπŸ›‘οΈ Security

President Trump says they’re looking into Binance hack as calls grow for crypto rules

The Block reports that Binance lost about $6.3 billion in a hack on 29 October 2026. Co-CEO Richard Teng said a compromised DevOps engineer's laptop gave access to a service container used to submit unauthorized withdrawals, and TRM Labs attributed the theft to North Korea's TraderTraitor. The rest covers reactions: President Trump said on 30 October he was looking into it, CFTC Chair Michael Selig and The Digital Chamber urged federal exchange rules, and senators disputed whether the Clarity Act would have allowed freezes.

BinanceTraderTraitorTRM Labs+5

Threat Intel, BlockchainΒ· Published and added 30 Oct 2026

galaxy.comarticleπŸ›‘οΈ Security

Weekly Research Brief: Binance Hacked; Hyperliquid on Wall Street; Prediction Markets’ Brazil β€˜Miss’

Galaxy Research's weekly brief of 30 October 2026 opens with the Binance hack: about $6.3b drained across 10 blockchains, attributed by TRM Labs and Chainalysis to TraderTraitor (Lazarus-linked). The attacker used EIP-7702 with 50 pre-signed authorizations on Ethereum, and Binance's preliminary finding is a compromised DevOps laptop reaching a service container. Tether and Circle froze about $2.6b. The other sections cover Hyperliquid and prediction markets.

EthereumBitcoinUSDC+2
TraderTraitorLazarus GroupEIP-7702+7

Blockchain, Threat IntelΒ· Added 30 Oct 2026

galaxy.comarticleπŸ›‘οΈ Security

Here We Go Again: Binance Loses $6.3b in Largest Robbery in History

Galaxy Research newsletter on the roughly $6.3b drained from Binance across 10 blockchains, which TRM Labs and Chainalysis attribute to TraderTraitor. On Ethereum the attacker used EIP-7702 to turn a hot wallet into a sweeper contract backed by 50 pre-signed authorizations. Binance's preliminary finding is a compromised DevOps laptop, with no wallet key leak. Tether, Circle, Ripple and World Liberty froze about $2.7b; Galaxy also assesses Binance's ability to cover losses.

EthereumBitcoinTether+3
TraderTraitorLazarus GroupBinance+7

Blockchain, Threat IntelΒ· Added 30 Oct 2026

bankless.comarticleπŸ›‘οΈ Security

Binance Just Suffered the Biggest Hack in Crypto History. What Now?

Bankless report on the Binance hack of $6.3 billion, which Binance's preliminary investigation traced to a compromised DevOps engineer's laptop with access to withdrawal-authorizing systems. TRM Labs found valid EIP-7702 authorizations on Ethereum, and nine other chains were drained through normally signed transactions. TRM and Chainalysis link it to North Korea's TraderTraitor. Tether and Circle froze about $3 billion, and Binance says SAFU and corporate cash will cover the losses.

EthereumBitcoinSolana+1
BinanceTraderTraitorDPRK+9

Blockchain, Threat IntelΒ· Published and added 30 Oct 2026

decrypt.coarticleπŸ›‘οΈ Security

Secret Service, FBI Join Binance Hack Probe as Circle Freezes $823 Million

Decrypt reports that Binance is working with the FBI and U.S. Secret Service after a theft of about $6.4 billion, which TRM Labs and Chainalysis attribute to North Korea's TraderTraitor group, also linked to the February 2025 Bybit hack. Circle froze about $823 million in USDC and Tether froze USDT, though the page gives differing Tether totals. Binance first suspected a compromised DevOps engineer laptop, found more breach points, and saw no evidence of leaked wallet private keys.

USDCTether
TraderTraitorBinanceBybit+10

Blockchain, Threat IntelΒ· Published and added 30 Oct 2026

21shares.comarticleπŸ›‘οΈ Security

Binance hack: the largest robbery in financial history

Binance hack analysis from 21shares on the 29 October 2026 theft of about $6.3 billion from Binance. Binance's initial finding points to a compromised DevOps engineer's laptop with no sign of a leaked wallet key. Chainalysis attributes it to North Korea with high confidence, and about $3 billion was frozen by Tether, Circle and the Arbitrum Security Council. The attacker also borrowed about $44.6 million of WETH on Aave. Binance says SAFU and company funds will cover all losses.

AaveTetherUSDC+2
BinanceLazarus GroupDPRK+7

Threat Intel, BlockchainΒ· Published and added 30 Oct 2026

theblock.coarticleπŸ›‘οΈ Security

Binance says it has $6.34 billion ready to cover hack losses

The Block reports that Binance says it has set aside $6.34 billion of its own funds to cover a hack it puts at about $6.3 billion. Co-CEO Richard Teng says the entry point was a compromised DevOps laptop and no private key leak was found. About $3 billion was frozen by Tether, Circle, Ripple and others, withdrawals are suspended, and TRM Labs sees parallels with North Korean attacks. Technical detail is thin.

BinanceTRM LabsTether+4

Blockchain, Threat IntelΒ· Published and added 30 Oct 2026

aikido.devarticleπŸ›‘οΈ Security

Binance hack: how a compromised developer laptop can lead to a $6.3 billion breach

Aikido Security's analysis of the roughly $6.3 billion Binance incident, which Binance says may have started with a compromised DevOps engineer's laptop. It sets out what is known (no private key leak, TraderTraitor attribution by TRM Labs and Chainalysis, additional breach points) and a plausible chain from laptop credentials to a withdrawal service. It compares the 2025 Bybit theft and lists mitigations.

npmVS CodeAWS+1
BinanceTraderTraitorLazarus Group+9

Blockchain, Threat Intel, Supply ChainΒ· Published and added 30 Oct 2026

decrypt.coarticleπŸ›‘οΈ Security

Binance Finds 'Additional Points of Breach' in $6.3B Hack

Binance estimated the October 2026 security incident at about $6.3 billion and said it found "additional points of breach" beyond the compromised DevOps engineer laptop it first named. The attacker reached a service container and submitted unauthorized withdrawals, with no wallet private key leak found. Arkham counted $6.01 billion taken from hot wallets on nine chains; Tether and Circle froze about $1.8 billion in stablecoins. About $3 billion is frozen onchain; withdrawals stay suspended.

TetherUSDCEthereum+2
BinanceArkhamRichard Teng+4

Threat Intel, BlockchainΒ· Published and added 30 Oct 2026

bleepingcomputer.comarticleπŸ›‘οΈ Security

Binance links $6.3 billion crypto theft to compromised DevOps laptop

Binance linked the theft of about $6.3 billion in crypto to a compromised DevOps engineer's laptop, which let attackers reach a service container and submit unauthorized withdrawals. TRM Labs and Chainalysis attribute the 29 October 2026 attack to North Korea's TraderTraitor. About $2.7 billion was frozen, and the attackers hold about $2.4 billion. Binance has found no evidence of a leaked wallet private key.

BinanceTetherCircle+3
TraderTraitorLazarus GroupAPT38+5

Blockchain, Threat IntelΒ· Published and added 30 Oct 2026

securityweek.comarticleπŸ›‘οΈ Security

$6 Billion Binance Hack Bears Hallmarks of North Korea’s TraderTraitor

The Binance hack of 29 October 2026 drained about $6.36 billion across ten blockchains in roughly 20 minutes, according to TRM Labs. TRM attributes it with high confidence to North Korea's TraderTraitor, citing a mixer-funded operator wallet, pre-signed transactions and overlap with the Bybit and KelpDAO hacks. Binance's early findings point to a compromised DevOps engineer laptop. Stablecoin issuers froze about $2.7 billion.

EthereumTornado CashTether+2
TraderTraitorLazarus GroupDPRK+11

Blockchain, Threat IntelΒ· Published and added 30 Oct 2026

en.cryptonomist.charticleπŸ›‘οΈ Security

Binance probes a suspected cryptocurrency breach: $6.3B affected, about $3B frozen

Binance disclosed a suspected security incident on 29 October 2026, estimating $6.3 billion affected and about $3 billion frozen. Binance's early finding is that a compromised DevOps engineer's laptop let an attacker reach a service container and submit unauthorized withdrawals, with no evidence of wallet private-key leakage. TRM Labs suspects TraderTraitor, the group tied to the February 2025 Bybit theft. Tether, Circle and Ripple froze funds.

TetherUSDC
BinanceTraderTraitorLazarus Group+7

Blockchain, Threat IntelΒ· Published and added 30 Oct 2026

x.comthreadπŸ›‘οΈ Security

Trail of Bits (@trailofbits) on X

Trail of Bits post on X summarising Binance's initial investigation into a security incident. Binance suggests a compromised DevOps engineer's laptop let an attacker reach a service container and submit unauthorized withdrawal transactions. The text says no evidence has been found of wallet private key compromise. The fetched text is truncated and gives no further detail.

BinanceTrail of Bits

Threat Intel, BlockchainΒ· Published and added 30 Oct 2026

opensourcemalware.comtalkπŸ›‘οΈ Security

The OpenSourceMalware Show #27

The OpenSourceMalware Show #27 (28 October 2026) is a podcast episode with transcript. It covers the Brotli decoder bug (CVE-2026-111497) vendored in C code across Python, Rust and PHP, and how to find copies via decode.c. It also covers the LiteLLM v1.107.0 compromise from an unrevoked admin key and a relaxed release control, and the MALFEX npm campaign hiding payloads in PNG-header files, plus an AGENTS.md lure and DPRK fake fonts in TTF/OTF.

BrotliLiteLLMnpm+3
CVE-2026-111497Node2ShellPHP2Shell+10

Supply Chain, Threat Intel, DevSecOpsΒ· Published and added 29 Oct 2026

arxiv.orgpaperπŸ›‘οΈ Security

Appreciation Teaches Unauthorized Execution: Self-Evolution Degrades the Robustness of Personal Agents to Indirect Prompt Injection

Paper studying whether self-evolution weakens personal agents' robustness to indirect prompt injection (IPI). A two-phase method has the user ask the agent to follow tasks in external content, then tests with a third-party injected task the user did not authorize. Against the non-evolving base agent, attack success rose 14.0 to 48.8 points under skill evolution and 3.0 to 39.3 points under reinforcement learning. Code is released.

indirect prompt injection

AI SecurityΒ· Added 31 Oct 2026, published 29 Oct 2026

ethresear.chpaperπŸ›‘οΈ Security

Secure Transaction Verification and Authorization (STVA)

Secure Transaction Verification and Authorization (STVA) is a draft Standards Track ERC (created October 2026) by @MATOBOYCRYPTO65. It proposes structured human-readable metadata and semantic intent for transactions and UserOperations, an independent simulation and verification interface, and scoped, revocable capability tokens for privileged operations. It targets blind signing and frontend supply-chain attacks, citing the February 2025 Bybit Safe multisig theft. The page gives only the abstract, motivation and core concepts.

EthereumSafe
BybitEIP-712ERC-1271+5

BlockchainΒ· Published and added 28 Oct 2026

github.blogarticle🧠 AI

GitHub Universe 2026: Evolving every layer of the stack

GitHub Universe 2026 recap in which GitHub presents its agentic software delivery vision and new Copilot products: the Copilot app, Copilot Fusion multi-model routing, Copilot projects, session continuity, automations that learn from outcomes, and code review Max. GitHub Enterprise customers can use all Copilot features without a separate subscription. GitHub also reports added capacity, a multi-region Azure migration and a redesigned Git architecture.

GitHub CopilotGitHubGit+4
MicrosoftJay ParikhCopilot Fusion+4

coding-agents, agents, ai-infraΒ· Added 29 Oct 2026, published 28 Oct 2026

securityweek.comarticleπŸ›‘οΈ Security

TLS Flaws Found in GoAhead Web Server Used in Hundreds of Millions of Devices

FoxIO disclosed three TLS flaws in the GoAhead embedded web server from EmbedThis, used on hundreds of millions of IoT, OT and network devices. CVE-2026-115881 (CVSS 9.1) leaves revoked client certificates accepted under mTLS with OpenSSL; CVE-2026-115884 (7.4) bypasses chain-length limits with LibreSSL; a third Mbed TLS bug (5.9, no CVE) weakens TLS 1.2 DHE. GoAhead 6.1.1 fixes the OpenSSL and Mbed TLS issues, and FoxIO released PoC code and fingerprints.

GoAheadOpenSSLMbed TLS+2
CVE-2026-115881CVE-2026-115884FoxIO+4

Hardware & RF, AppSecΒ· Published and added 28 Oct 2026

arxiv.orgpaper🧠 AI

FeatureArc: Benchmarking Coding Agents on Evolving Features

FeatureArc is a benchmark for coding agents on evolving features, built by a pipeline that mines repository histories into multi-stage tasks across four programming languages. It has 73 manually reviewed tasks of 2 to 10 stages, validated with fail-to-pass and pass-to-pass tests. Across 10 agents, pooled success drops from 62.7% at the first stage to 28.5% for the whole feature. Inherited defects appear in 72.9% of failed later stages.

LLM
FeatureArcTencentHunyuan+1

evals, coding-agents, researchΒ· Added 28 Oct 2026, published 27 Oct 2026

securityweek.comarticleπŸ›‘οΈ Security

Vendor Key Leak Puts Data of 238,000 Denver Health Plan Members at Risk

Denver Health Medical Plan is notifying 238,083 members that vendor SquareML leaked an access key on a public website. The key was online from 27 June 2025 until it was revoked on 18 August 2026 and gave read-only access to a database of names, dates of birth, health metrics, eligibility and admission data. DHMP found no evidence of misuse and reset passwords for all accounts with access.

Denver Health Medical PlanSquareMLHealth First Colorado+2

Threat IntelΒ· Published and added 27 Oct 2026

x.comthreadπŸ›‘οΈ Security

David Gutierrez | Eng (@david_x_ai) on X

Thread on two October 2026 npm compromises: @subql/common 5.8.3 (5 October), published through GitHub Actions OIDC Trusted Publishing after the attacker tampered with the publish workflow, and tensorlake 0.5.144 (8 October), published by the project's own release workflow with valid provenance for about 1 hour 40 minutes. Both ran on install and stole cloud, GitHub, Kubernetes, Vault, SSH and .env credentials; tensorlake also took AI tool and MCP config. Its payload installed a gh-token-monitor service that deletes the home directory if the stolen GitHub token is revoked, so remove it before rotating.

npmGitHub ActionsGitHub+3
@subql/commontensorlakegh-token-monitor+3

Supply Chain, DevSecOpsΒ· Added 27 Oct 2026, published 26 Oct 2026

arxiv.orgpaper🧠 AI

IntentDelta: A Human-Constructed Benchmark for Evolving User Intent

IntentDelta is a human-constructed multi-turn benchmark of 638 trajectories and 3,766 turns across WebShop and TravelPlanner. Each turn pairs a user utterance with a cumulative intention state and human reference actions. Across eleven agents, average Intention Recovery is 33.1% in TravelPlanner and 44.7% in WebShop, and average Action Score is 31.6% and 41.0%. Agents lose earlier requirements and confuse soft preferences with must-have constraints.

LLM
IntentDeltaWebShopTravelPlanner+1

evals, agents, researchΒ· Added 27 Oct 2026, published 24 Oct 2026

x.comthreadπŸ›‘οΈ Security

The Hacker News (@TheHackersNews) on X

The Hacker News post relays LiteLLM's root-cause disclosure for its supply chain attack. A master key leaked on GitHub in May was never revoked. Attackers used it to reach the Agent Platform, steal a GitHub owner token, and publish a malicious PyPI release in October. A reply says the agent tool held the org owner's personal CLI login, which skips GitHub's 2FA re-check.

LiteLLMPyPIGitHub
CWE-798

Supply Chain, DevSecOps, Identity & AccessΒ· Added 26 Oct 2026, published 24 Oct 2026

unit42.paloaltonetworks.comadvisoryπŸ›‘οΈ Security

EMudhra Intermediate CA Certificates Compromised

Unit 42 advisory on the compromise of eight intermediate CA private keys belonging to Indian CA eMudhra Limited. The keys were generated with insufficient randomness, so attackers could derive them from the public certificates and sign rogue TLS certificates for MitM attacks. eMudhra has revoked the certificates and browser and OS vendors have untrusted them. Interim guidance covers PKI and trust store audits, reissuance, strict CRL/OCSP checking, log hunting, CT monitoring and CAA records.

Prisma BrowserJavaDocker+2
eMudhraUnit 42Palo Alto Networks Next-Gen Trust Security+1

Threat Intel, Blue Team & DetectionΒ· Added 30 Oct 2026, published 22 Oct 2026

huntress.comarticleπŸ›‘οΈ Security

Unboxing Phishing Kits: Huntress Examines 3 Real Kits

Recap of a Huntress Tradecraft Tuesday session (October 2026) on phishing kits that target Microsoft 365 tokens, including EvilTokens (a device-code phishing campaign abusing Railway) and the Kali365 control panel. It explains device-code phishing, adversary-in-the-middle and token replay. It lists detections (unfamiliar cloud sign-ins, device-code auth, new devices) and response steps (revoke refresh tokens, remove rogue devices, block device-code flow).

Microsoft 365Microsoft Entra IDHuntress+1
EvilTokensKali365device-code phishing+4

Threat Intel, Blue Team & Detection, Identity & AccessΒ· Published and added 22 Oct 2026

aikido.devarticleπŸ›‘οΈ Security

eMudhra CA compromise: eight private keys, and one of them is 4

eMudhra's eight intermediate CA certificates, chaining to emSign ECC Root CA - G3, had private keys below 256 (one is 4), disclosed on 17 October 2026 by a researcher known as Wayne. Browsers and Microsoft blocked them via revocation pushes, but curl, git, containers, language runtimes and IoT may still trust them. The post gives patched ca-certificates versions, test commands and the timeline.

cURLDockermacOS+3
eMudhraemSign ECC Root CA - G3Wayne+6

Supply Chain, DevSecOpsΒ· Published and added 22 Oct 2026

oracle.comadvisoryπŸ›‘οΈ Security

Oracle Security Alert Advisory – CVE-2026-91403 (eMudhra cert revocation)

Oracle Security Alert for CVE-2026-91403 ships Java SE updates that distrust a set of intermediate CA certificates tied to eMudhra. It contains one new patch for Oracle Java SE, described as remotely exploitable without authentication. Customers are told to update Java or apply mitigations listed at ops.java/emudhra and in the Patch Availability Document. No reporter is credited and the page gives no CVSS score or technical detail.

JavaCVSS
CVE-2026-91403eMudhra

Cloud & Infra, AppSecΒ· Added 23 Oct 2026, published 21 Oct 2026

blog.railway.comadvisoryπŸ›‘οΈ Security

Railway Security Disclosure

Railway's security disclosure of 21 October 2026 says an unauthorized third party may have accessed encrypted customer environment variables and decrypted some of them. Railway links it to a series of attacks on platform companies aimed at compromising AI inference providers, and has asked upstream providers to revoke keys. It tells all users to roll every environment variable, API key and registry credential, and it paused all deployments. The notice gives no technical details of the intrusion.

Railway
Railway security incident October 2026

Threat Intel, DevSecOpsΒ· Added 22 Oct 2026, published 21 Oct 2026

endorlabs.comarticleπŸ›‘οΈ Security

eMudhra CA Compromise: The Trust Stores You Didn't Patch | Endor Labs

The eMudhra CA compromise (disclosed 20 October 2026) involves eight intermediate CA certificates whose private keys can be derived from the certificates themselves. They chain to AAA Certificate Services and three other roots. Browsers block them, but non-browser clients skip revocation and CT checks. Endor Labs finds 25 popular packages with frozen bundled root stores, 11 already patched, and gives a fingerprint-scanning script, test hosts and patch steps.

Endor LabsPythoncertifi+6
CVE-2026-111943eMudhraAAA Certificate Services+9

Supply Chain, AppSec, DevSecOpsΒ· Published and added 21 Oct 2026

coindesk.comarticleπŸ›‘οΈ Security

A company that vouches for websites had secret keys so weak that 8 of them were cracked

eMudhra, an Indian certificate authority with over 60 million issued certificates, had eight CA private keys that were tiny integers such as 4, 56 and 237. Researcher Scott Helme recovered all eight from public certificates in under a second, and the keys were exposed from April 2025. Mozilla, Chrome, Apple and Microsoft blocked them between 17 and 20 October 2026, but curl, Python, Node.js, Go and Java accepted a forged Stripe API certificate because they skip revocation checks.

cURLPythonNode.js+3
eMudhraSectigoScott Helme+5

otherΒ· Published and added 21 Oct 2026

blog.cloudflare.comwriteupπŸ›‘οΈ Security

8 eMudhra certificates are compromised. Here's what you need to know

Cloudflare's incident post on the compromise of private keys for eight eMudhra intermediate certificates, disclosed 19 October 2026. The weak keys let an attacker forge certificates for any domain. It describes Cloudflare's patching of trust stores, WARP and cloudflared, tells developers to update the underlying TLS verifier and not disable verification, and points to Dennis Jackson's test site and revocation systems such as CRLSets and upki.

CloudflareCloudflare TunnelWARP+2
eMudhraDennis Jacksonupki+1

Cloud & Infra, Blue Team & DetectionΒ· Added 29 Oct 2026, published 20 Oct 2026

huggingface.cowriteup🧠 AI

ALTK-Evolve: How learned memory tops the τ²-bench leaderboard on customer service tasks without a frontier model

ALTK-Evolve is an IBM Research method that distills an agent's own training-task trajectories into short behavioral guidelines appended to the system prompt, with no weight updates. On τ²-bench, a stock ReAct agent with gpt-4.1 reaches 67.5% Pass^4 on retail, and gpt-5.5 reaches 82% on airline. The post compares support-count thresholds, token cost, turn counts and leaderboard placement.

ALTK-Evolveτ²-benchReAct+1
IBMSierraGPT-4.1+4

agents, evals, promptingΒ· Published and added 20 Oct 2026

securityweek.comarticleπŸ›‘οΈ Security

Hackers Steal 1.8 Million Records From Nikkei

Nikkei disclosed a breach of 1.83 million records from a cloud service it uses for system development, accessed on 7 October 2026 with previously compromised credentials. About 150,000 records held names, addresses, emails, phone numbers and birth dates, and 120,000 held email addresses only. Nikkei revoked the credentials and notified regulators. It found no misuse as of 16 October, and it had disclosed an earlier Slack account breach less than a year before.

NikkeiFinancial Times

Threat IntelΒ· Published and added 20 Oct 2026

arxiv.orgpaper🧠 AI

MemT-1.0: Towards Memory-Native Evolution

MemT-1.0 is a memory foundation model architecture that builds memory into model computation. It builds on Metis, uses Gated DeltaNet-2 for token-level memory updates within each turn, and adds Mixture-of-Storage to scale capacity. Training used 34B tokens in continued pre-training and post-training. MemT-1.0-Nano and MemT-1.0-Flash beat parametric memory baselines and stay competitive with retrieval methods. At 128K history, Nano gives about 1.86x inference speedup and 66x less persistent storage than a full-context KV cache.

Gated DeltaNet-2Mixture-of-StorageKV cache+1
MemT-1.0MemT-1.0-NanoMemT-1.0-Flash+1

research, models, agentsΒ· Added 21 Oct 2026, published 19 Oct 2026

arxiv.orgpaper🧠 AI

Mitigating Reward Hacking in Preference Reward Models with Evolving Penalties

Dynalize is a method that reduces reward hacking during online RL training against preference reward models. Periodically and asynchronously, a language model reads training rollouts and writes penalties for undesirable behaviors the reward model seems to reward, and a judge applies them to every rollout. In clinical counseling and creative writing it sustains longer training and improves external benchmarks, and the gains hold in multi-turn training and after reward model retraining.

LLM
DynalizearXiv

training, research, ai-safetyΒ· Added 21 Oct 2026, published 19 Oct 2026

arxiv.orgpaper🧠 AI

How Effectively Do Security Skills Guide Coding Agents in Vulnerability Detection in Linux Kernel?

The paper is a large-scale study of 37 security skills run through Claude Code with four LLMs on 100 real Linux kernel vulnerabilities, 14,800 runs in all. Only 24.4 percent of runs found the correct function and 15.7 percent classified the vulnerability correctly. It proposes SKILL-SIF, which evolves skills from failure patterns and raised fully correct detection on 100 unseen vulnerabilities from 2 to 16 percent.

Claude CodeLinuxLLM
SKILL-SIF

evals, coding-agents, promptingΒ· Added 21 Oct 2026, published 19 Oct 2026

arxiv.orgpaperπŸ›‘οΈ Security

How Effectively Do Security Skills Guide Coding Agents in Vulnerability Detection in Linux Kernel?

Empirical study of 37 security skills run through Claude Code with four LLMs on 100 real Linux kernel vulnerabilities across ten classes (14,800 runs). 24.4% of runs found the correct function, 18.3% the function and statement, and 15.7% the vulnerability class. The authors propose SKILL-SIF, which evolves skills from failures and raised fully correct detection on 100 unseen vulnerabilities from 2% to 16%.

Claude CodeLinux kernelLLM
SKILL-SIF

AI Security, AppSecΒ· Added 21 Oct 2026, published 19 Oct 2026

huggingface.copaper🧠 AI

Self-evolving Agent Harness

Self-evolving Agent Harness is a 66-page survey and theory paper on revising an agent's scaffolding (prompts, memory, workflows, tools, code, plugins) while model weights stay fixed. It frames search as evolutionary optimization and uses continual PAC-learning to derive a generalization bound, a retention bound and a high-probability admission rule with fresh validation data. It separates selection, activation, attribution and improvement as evaluation claims, and covers applications and governance. Code is at github.com/zgcagi/Harness-RSI.

LLM
Harness-RSIHugging Face

agents, research, evalsΒ· Added 19 Oct 2026

securitylabs.datadoghq.comwriteup🧠 AI

Putting models to the secure coding test: Security rules

Datadog Security Labs tests whether Project CodeGuard v1.4.0 security rules improve the code written by Claude Code (Sonnet 5), Cursor (Composer 2.5) and Codex (GPT 5.5). Each model built the same document portal app once. Only GPT 5.5 closed the cross-user document IDOR and improved clearly. Sonnet 5 and Composer 2.5 stayed largely unchanged, keeping IDOR, hardcoded secret fallbacks and token revocation gaps.

Claude CodeCursorCodex+9
DatadogClaude Sonnet 5Composer 2.5+1

coding-agents, evals, promptingΒ· Added 29 Oct 2026, published 19 Oct 2026

securitylabs.datadoghq.comwriteupπŸ›‘οΈ Security

Putting models to the secure coding test: Security rules

Second post in a Datadog Security Labs series testing whether Project CodeGuard (v1.4.0) security rules improve code written by coding agents. Sonnet 5 (Claude Code), Composer 2.5 (Cursor) and GPT 5.5 (Codex) each built the same document portal, and the output was audited with Datadog Code Security and a Sonnet 5 code review. Only GPT 5.5 closed the cross-user IDOR; Sonnet 5 and Composer 2.5 stayed largely unchanged, including hardcoded JWT secret fallbacks and incomplete revocation.

Claude CodeCursorCodex+3
Project CodeGuardIDORCSRF+3

AI Security, AppSecΒ· Added 29 Oct 2026, published 19 Oct 2026

decrypt.coarticle🧠 AI

TokenMaxxing: How to Get the Most Out of Your AI Models

Decrypt guide to "tokenmaxxing" reframed as maximizing value per token rather than burning tokens. It gives tiered tips: pick smaller models for simple jobs, write one clear prompt, start new chats, cap output length, and use free models (beginner); Claude Code commands like /model, /clear, /compact, /context, /mcp, /effort and plan mode (intermediate); and prompt caching via cache_control, 50% batch discounts, and agent budget limits like --max-budget-usd (advanced). Includes context on Meta, Amazon and Uber token leaderboards and spend, a worked invoice-processing example where routing cut cost from $30 to $3.90, and the Jevons paradox in falling token prices.

Claude CodeClaudeChatGPT+8
DecryptAnthropicOpenAI+14

ai-infra, coding-agents, ai-policyΒ· Published and added 17 Oct 2026

en.cryptonomist.charticle🧠 AI

AI generated viral genomes yield 16 viable phages, prompting safety concerns

Reports a Science study from Stanford University and the Arc Institute in which Evo, a genome language model trained on sequences from across the domains of life, generated thousands of bacteriophage genome designs targeting E. coli. About 300 were built and tested, and 16 produced viable viruses. A mixture of the AI-designed phages overcame resistance in some E. coli strains where a comparable mixture of natural phages did not. The article also covers a Johns Hopkins Center for Health Security commentary saying governance lags behind generative genome design, and a more cautious biosafety assessment from an outside expert.

Evo
Stanford UniversityArc InstituteScience+4

research, ai-safetyΒ· Published and added 16 Oct 2026

arxiv.orgpaper🧠 AI

GLOBE: Evolving Structured Agentic Memory for E-commerce Product Optimization in Generative Search

GLOBE is an agentic framework that optimizes e-commerce product listings for generative search by storing strategies in an evolving handbook of explicit, field-scoped behaviors. An Editor applies behaviors and gets rank-based feedback, a Reflector turns it into updates, a Moderator removes redundant proposals, and a Curator commits changes. Harmful behaviors are deactivated. Structured behaviors beat monolithic prompt optimization and transfer to unseen product categories.

LLM
GLOBEarXiv

agents, researchΒ· Added 20 Oct 2026, published 16 Oct 2026

securityweek.comarticleπŸ›‘οΈ Security

F5, Juniper Networks Patch Critical Vulnerabilities

SecurityWeek reports F5 fixed 11 NGINX vulnerabilities. Three are critical: CVE-2026-15176 and CVE-2026-93479 are unauthenticated buffer overflows via crafted HTTP requests, and CVE-2026-97205 is memory corruption. All three can lead to RCE if ASLR is disabled or bypassed. Juniper Networks published 17 advisories covering over 150 flaws in Junos OS and Junos OS Evolved, plus four in Cloud-Native Router. About two dozen are critical, with root code execution, device takeover, kernel panic and an off-by-one buffer overflow. Neither vendor reports exploitation in the wild.

nginxJunos OSJunos OS Evolved+1
CVE-2026-15176CVE-2026-93479CVE-2026-97205+2

AppSec, Cloud & InfraΒ· Published and added 16 Oct 2026

opensourcemalware.comwriteupπŸ›‘οΈ Security

GhostAction Copycats PolinRider by Hiding a Stealer in a Fake Font File

Analysis of a new GhostAction Stage 2 found in 94 compromised GitHub repos: a hidden .vscode/tasks.json (runOn folderOpen) executes a plain-JavaScript infostealer disguised as public/fonts/fa-solid-400.woff2. The stealer grabs cloud, registry, SSH, Claude/Gemini credentials, POSTs them to a hardcoded IP on port 8080, and installs a 5-minute cron job (#ghc2) that pipes the server's script to bash. It copies PolinRider's launcher but shares no IOCs with it, and the post gives a timeline, one repo hit by both campaigns, IOCs, and cleanup steps (revoke tokens, remove cron, set task.allowAutomaticTasks off).

VS CodeGitHub ActionsGitHub+3
GhostActionPolinRiderDPRK+10

Threat Intel, Supply Chain, Malware & REΒ· Added 15 Oct 2026, published 16 Oct 2026

github.blogarticleπŸ›‘οΈ Security

Staying Evergreen: How we automated secure access and repo hygiene in a shared org

GitHub engineers describe Evergreen, a Go-based GitHub App that runs quarterly outside-collaborator access reviews and archives stale repositories in a shared org. It uses scheduled Actions with repository_dispatch, JSON state files in git as the audit trail, checklist issues (IssueOps) closed to trigger revocation, and custom properties for ownership and exemptions. Design principles include fail-secure/fail-closed revocation, never deleting data, dry-run rollout, Splunk logging, and proactive pacing against GitHub's secondary rate limits.

GitHubGitHub ActionsGo+2
EvergreenIssueOpsfail secure+2

DevSecOps, Identity & AccessΒ· Published and added 15 Oct 2026

eprint.iacr.orgpaperπŸ›‘οΈ Security

Threshold Post-Quantum E-Voting from Lattice Commitments

T-EVOLVE extends the lattice-commitment e-voting protocol EVOLVE to threshold tallying. Votes are Shamir-shared, so any t of n authorities can finish the tally. Privacy holds against t-1 colluding authorities and verifiability holds even if all of them collude. The voter proves her committed shares form a Reed-Solomon codeword whose value at zero is a valid vote. Complaints about ciphertexts that do not decrypt are publicly verifiable, using an LWE-instantiable encryption scheme with publicly verifiable opening. Privacy, verifiability, individual verifiability and robustness are proven in the classical random oracle model, and the verifiability reduction is not tight. In a C implementation, a yes/no ballot for four authorities takes 109 KB, and an authority aggregates 10^4 ballots in about three minutes on one core.

C
T-EVOLVEEVOLVEShamir secret sharing+2

otherΒ· Added 17 Oct 2026, published 15 Oct 2026

snyk.ioarticle🧠 AI

VulnBump: Can AI Agents Fix Vulnerable Dependencies Without Breaking the Build?

Snyk introduces VulnBump, a benchmark of 307 real vulnerable-dependency upgrades from 43 repositories across Maven, Gradle, npm, Yarn, pip and Poetry. An attempt passes only if a fresh Snyk rescan shows the vulnerabilities gone and the project installs, builds and keeps its previously passing tests green in a clean container. On Claude Sonnet 5.5, Snyk's Evo Remediation Agent passed 90.2%, Claude Code with Snyk scan and target version 77.5%, unaided Claude Code 59.6%, and a non-AI package-manager bump baseline 46.9%. Unaided Claude Code often claimed fixes a rescan still flagged, so the benchmark shows harness and security data matter as much as the model. The harness and 43 cases are open-sourced.

Claude CodeSnyk Open SourceMaven+6
SnykVulnBumpVulnBench+6

evals, coding-agentsΒ· Published and added 15 Oct 2026

snyk.ioarticleπŸ›‘οΈ Security

VulnBump: Can AI Agents Fix Vulnerable Dependencies Without Breaking the Build?

Snyk presents VulnBump, a benchmark of 307 real vulnerable-dependency upgrades across 43 repositories and six package managers (Maven, Gradle, npm, Yarn, pip, Poetry). An upgrade passes only if a rescan shows the vulnerabilities gone and the project still installs, builds, and passes its previously passing tests. On Claude Sonnet 5.5, Snyk's Evo Remediation Agent passed 90.2%, Claude Code with Snyk scan 77.5%, Claude Code unaided 59.6%, and a plain package-manager bump 46.9%. Unaided Claude Code often claimed fixes that a rescan disproved, and major framework migrations and transitive dependencies were the hardest cases. The harness and the 43 cases are open-sourced.

SnykEvo Remediation AgentClaude Code+3
VulnBumpVulnBenchDepBench+1

AI Security, Supply ChainΒ· Published and added 15 Oct 2026

arxiv.orgpaper🧠 AI

Recursive Reward Improvement via Adversarial Meta-Prompt Evolution

Introduces Recursive Reward Improvement (RRI) for recursive self-improvement on open-ended expert tasks without answer keys, where a proposer writes both a task and a rubric. Before the solver trains, an adversary writes a rubric-gaming but useless answer, the same model writes an honest answer without seeing the rubric, and up to two repair steps revise the rubric until the honest answer wins. Exploits found also rewrite future task and rubric generation. On HealthBench Professional it beats a Generation Only baseline by 10.9 points at 9B and 12.4 at 27B (10.3 with no external model), and the 27B model transfers to PRBench Hard (+6.6) and ProfBench (+8.0).

LLM
HealthBench ProfessionalPRBench HardProfBench+2

research, trainingΒ· Added 17 Oct 2026, published 15 Oct 2026

huntress.comarticleπŸ›‘οΈ Security

Passkeys Raise the Bar. Here’s Where Attackers Can Still Get In.

Huntress post on where accounts remain exposed after a passkey rollout. It covers passkey-themed social engineering, the EvilTokens device-code phishing campaign (344 organizations in about 16 days), weak enrollment and recovery, SpecterOps research on Windows WebAuthn hooking on compromised endpoints, and the trade-offs between synced and device-bound passkeys. It closes with ITDR-style post-login monitoring, containment steps (revoke sessions, remove rogue authenticators, hunt inbox rules, OAuth grants, forwarding), and a five-question deployment checklist.

WebAuthnFIDO2Windows+2
EvilTokensdevice code phishingpasskey-themed phishing+4

Identity & Access, Blue Team & Detection, Threat IntelΒ· Published and added 14 Oct 2026

securityweek.comarticleπŸ›‘οΈ Security

US Arrests Two in Crackdown on North Korean IT Worker Laptop Farms

SecurityWeek reports on the arrest of two US nationals (Michael James Bostelle, Juan Cecilio Pena Carrillo) for running laptop farms that helped North Korean IT workers get remote jobs at over 70 US companies, generating more than $3M between 2019 and July 2025. It covers the FBI's Operation Lonestar Roundup (40 Texas locations searched, 187 laptops and KVMs seized). It also covers CrowdStrike's Famous Chollima analysis of shipping-diversion tricks and evolving evasion (hidden RMM tools, spoofed device hashes), and the FBI's 'identity in depth' guidance for spotting ITWs across the employment lifecycle.

CrowdStrikeVPN
Famous ChollimaDPRKOperation Lonestar Roundup+5

Threat Intel, Blue Team & DetectionΒ· Published and added 14 Oct 2026

latent.spacearticle🧠 AI

[AINews] AI Engineer NYC: Gemini 4 Argon, Vals AI Benchmarks

AINews issue for 10/12–10/13/2026 covering the GA of Google's Gemini 4 Argon (pricing, Cursor/Devin/Copilot availability, CursorBench 52.8%, Artificial Analysis Intelligence Index parity with GPT-6 Astra at lower cost) and Meta's Muse Wonder, with independent evals and reliability caveats. It also summarises decision-model benchmarks (AA Decision Index, Jev 1.13, Nimble-v4), Fish Audio's open-weight Drama 3 TTS, agent-harness and agentic-RL research (Evo2, ReHarness, LongHarness, ThunderSync), Anthropic's move off Git, and safety items such as Argon's monitor evasion. It also notes AI Engineer NYC keynotes on AI x Finance. Useful as a curated, caveat-aware snapshot of the frontier-model landscape.

GeminiGoogle AntigravityCursor+14
Gemini 4 ArgonGoogleGoogle DeepMind+43

models, evals, agentsΒ· Published and added 14 Oct 2026

arxiv.orgpaper🧠 AI

ProAS: A Progress-Aware Advantage Shaping Framework for Multi-Turn Agentic Reinforcement Learning

ProAS (Progress-Aware Advantage Shaping) is a plug-and-play framework for multi-turn agentic RL that fixes the uniform credit assignment problem in critic-free algorithms like GRPO, where trajectory-level advantages are broadcast to every action. It uses a dynamically evolving Agentic Reference Graph with Masking Unreasonable Incentives and Asymmetric Penalty Redistribution to give finer-grained credit without a critic. On Qwen2.5-1.5B/7B-Instruct it reports success rates up to 91.9%/94.3% on ALFWorld and 66.7%/70.8% on WebShop, with up to 2.44x faster training convergence and robustness to imperfect or distilled ARGs.

GRPOQwen
ProASAgentic Reference GraphALFWorld+4

training, agents, researchΒ· Added 14 Oct 2026, published 13 Oct 2026

theblock.coarticleπŸ›‘οΈ Security

Ledger confirms hardware implant in CryptoBilis reseller units as MediaMarkt suspends sales in Europe

Ledger's Donjon team confirmed that Ledger signers sold by the reseller CryptoBilis contain an unauthorized implant. It captures the 24-word recovery phrase during setup and sends it over LTE using a provisioned eSIM. It does not touch the Secure Element, the OS, or transaction signing. Onchain researchers estimate losses above $86M, but Ledger says that figure is unverified. MediaMarkt Saturn halted European Ledger sales after a tampered unit was reported in Germany, and Trezor and OneKey tightened or revoked reseller authorizations. The case shows a reseller-channel tampering risk for hardware wallets.

LedgerTrezorOneKey
CryptoBilisLedger DonjonMediaMarkt Saturn+2

Blockchain, Hardware & RF, Threat IntelΒ· Published and added 13 Oct 2026

arxiv.orgpaper🧠 AI

EmoGenie: Benchmarking and Evaluating Expressed and Evoked Emotion in T2I Generation with MLLMs

EmoGenie is a benchmark for emotions expressed and evoked by text-to-image outputs. It has 6,400 images from eight T2I models using 800 emotion-targeted prompts. It includes EmoGenie-QA (90,000 closed-set and 1,057 open-ended questions) and EmoGenie-Assess (13,175 mean opinion scores for valence, arousal, dominance and aesthetic quality). The authors also propose EmoProbe, which pairs emotion QA with response-aligned, dimension-specific regression heads. They report that it beats baselines on emotion QA and correlates better with human ratings. The work shows how reliably MLLMs interpret affect in generated images.

multimodal large language modeltext-to-image
EmoGenieEmoGenie-QAEmoGenie-Assess+2

evals, multimodal, researchΒ· Added 14 Oct 2026, published 13 Oct 2026

securityweek.comarticleπŸ›‘οΈ Security

XWorm RAT Evolves Into Surveillance Platform

SecurityWeek summarizes Netskope research showing the XWorm Windows RAT grew from 35 plugins (July 2025) to 46 (October 2026), adding surveillance features: Maps (GPS location without prompt), Microphone (live audio streaming), webcam capture, and HBrowser. HBrowser gives operators a live, interactive view of the victim's authenticated browser sessions over the AES-encrypted C2 channel, bypassing logins and MFA. Activity rose after the builder was cracked in June 2025 and the 6.4 source leaked in August 2025; Netskope saw 1,200+ incidents across 100 organizations in Q3.

Windows
XWormNetskopeHBrowser+2

Threat Intel, Malware & REΒ· Published and added 13 Oct 2026

aws.amazon.comarticleπŸ›‘οΈ Security

Quantum-resistant signing on AWS, Part 1: Build a software supply chain with ML-DSA

Part 1 of an AWS series showing how to build a quantum-resistant software supply chain with ML-DSA: ML-DSA keys in AWS KMS, an ML-DSA root/subordinate CA hierarchy in AWS Private CA, and IAM Roles Anywhere trust anchors for external build agents. It covers signing binaries (EXTERNAL_MU with OpenSSL 3.5), container images as OCI referrers via ORAS, and SBOMs wrapped in in-toto Statements bound to the artifact digest, plus verification, admission control, trust-store distribution, revocation and SCP guardrails. It frames the "forge later" threat and maps CNSA 2.0 (ML-DSA-87 only for NSS) and EU CRA deadlines to actions.

AWSAWS KMSAWS Private CA+3
ML-DSANIST FIPS 204CNSA 2.0+8

Supply Chain, DevSecOps, Cloud & InfraΒ· Published and added 12 Oct 2026

arxiv.orgpaperπŸ›‘οΈ Security

StableJudge: A Lightweight Adversarial Defense Framework for LLM-as-a-Judge

StableJudge is a defense framework against score-hijacking prompt injection in LLM-as-a-Judge pipelines (e.g. paper review, recommendation). A compact 0.13B "Sentry" model localizes injected spans and sanitizes them before the judge scores the content, trained on adversarially evolving attacks, and deterministic fine-grained scores are derived from the judge's output distribution. It improves score recovery under attack on held-out scientific abstracts and out-of-distribution attack variants while preserving general performance, decoupling safety from judge model scale. Code and data are promised but not yet released.

LLM
prompt injectionindirect prompt injection

AI SecurityΒ· Added 15 Oct 2026, published 12 Oct 2026

interconnects.aiarticle🧠 AI

I expect rapid progress but not towards general superintelligence

Nathan Lambert argues that near-term AI progress will come mostly from automating infrastructure and engineering work (superhuman distributed GPU engineers, end-to-end optimization of training and inference), not from models becoming broadly superhuman researchers. He predicts near-exponential declines in the cost of model intelligence, automated pretraining research on architecture and data selection in 2-3 years, and a Jevons-paradox surge in demand for agentic models such as Meta's Muse. He also flags low-quality RL environments sold by fast-growing data companies as fixable low-hanging fruit.

GPU
Nathan LambertInterconnectsIlya Sutskever+2

research, training, inference, ai-policyΒ· Published and added 9 Oct 2026

arxiv.orgpaper🧠 AI

Images Talk Back: Guiding MMDiTs by Suppressing Image-to-Text Feedback

The paper studies joint attention in multimodal diffusion transformers (MMDiTs), where the evolving image updates text representations. It finds that this image-to-text (I2T) feedback tends to preserve objects and attributes already in the image, errors included, and that suppressing it improves generation. It proposes Semantic Feedback Guidance (SFG), which guides away from an auxiliary branch that strengthens I2T feedback and weakens text-to-image conditioning, and iSFG, an in-place variant that needs no extra model evaluations. SFG runs for the first 20% of sampling steps (about 1.2x base NFE without CFG). Across three backbones, adding SFG to CFG on FLUX raises GenEval from 0.6661 to 0.7479, and iSFG on SD3.5M adds +1.33 HPSv2.1 over CFG at zero extra NFE. The control also improves VBench scores for text-to-video and image-to-video.

diffusion modelTransformerFLUX+1
SD3.5MGenEvalHPSv2.1+4

research, multimodalΒ· Added 12 Oct 2026, published 9 Oct 2026

x.comadvisoryπŸ›‘οΈ Security

β™«Whyβ™₯Notβ™ͺ (@Python_s_) on X

Alert summarizing Socket Threat Research's disclosure that a malicious version of the Tensorlake npm SDK (an AI-agent infrastructure package) was published with a preinstall hook delivering Shai-Hulud/ChainDrop credential-stealing malware. The payload targets GitHub, npm, AWS, Kubernetes and Vault credentials, persists via a gh-token-monitor mechanism, and spreads through compromised publishing accounts. It lists IOCs (lib/setup.mjs, lib/Math_Symbol.js) and warns to remove the persistence before revoking stolen GitHub tokens, since revocation triggers a destructive routine. Socket detected the release about 11 minutes after publication.

npmTensorlakeSocket+3
Shai-HuludChainDropgh-token-monitor+2

Supply Chain, Threat IntelΒ· Added 11 Oct 2026, published 9 Oct 2026

securityweek.comarticleπŸ›‘οΈ Security

Google Domains Impacted by Recent ccTLD Hijacks

Google disclosed that attackers hijacked the authoritative DNS of three ccTLDs (.gh for Ghana, .sl for Sierra Leone, .as for American Samoa) and used control of those domains to obtain unauthorized HTTPS certificates covering several Google domains and domains of other organizations. Google blocked the fraudulent certificates in Chrome, worked with issuing CAs to revoke them, and used Certificate Transparency log analysis to find other affected brands, some of which it proactively notified. Google is urging domain owners to monitor CT logs and publish restrictive CAA records after regaining DNS control, since CAs may cache and reuse prior domain-control-validation checks, letting attackers mint new certs even after the hijack ends.

Chrome
.gh ccTLD.sl ccTLD.as ccTLD+3

Threat IntelΒ· Published and added 9 Oct 2026

galaxy.comarticle🧠 AI

Weekly Research Brief: AI Governance, Shielded Bitcoin, and Mention Markets

Galaxy Research's weekly brief devotes its lead section to AI governance: it recaps Sam Altman's and Dario Amodei's UN Security Council remarks calling for international AI safety coordination, contrasts a disclosed OpenAI agent incident (an agent accessed non-public Medicare data in Australia, reported three months late) with Anthropic's 950-agent DNA-database research run, and cites concrete agent-economy data β€” OpenRouter's agent token usage overtaking human usage and growing ~14x since February, Cloudflare's agent-traffic crossover, and Vercel AI Gateway data showing a 23% monthly price drop and open-weight models rising to 56% of token volume. The analysis argues that as inference gets cheaper and open-weight models gain share, AI governance's hard problem shifts from controlling frontier labs to controlling where cheap, widely available agents can act. The rest of the newsletter covers an unrelated Bitcoin privacy protocol proposal and CFTC guidance on prediction "mention markets."

OpenRouterCloudflareVercel AI Gateway+1
Sam AltmanDario AmodeiOpenAI+7

ai-policyΒ· Added 8 Oct 2026

arxiv.orgpaper🧠 AI

Cited but Not Consulted: A Counterfactual Audit of Legal Chain-of-Thought Faithfulness

The paper tests whether LLM legal explanations reflect what drives the verdict. Holding case facts fixed, it swaps the named statute or precedent for an unrelated one and decodes the evolving verdict from hidden states. Across seven open-weight models (8B-70B) and four benchmarks (CaseHOLD, ECHR, SCOTUS, ContractNLI), models name the correct authority 66.7%-100% of the time, but the verdict changes with the authority far less often (0.0%-21.7% on CaseHOLD, 30.0%-76.7% on ECHR and SCOTUS, 43.3%-50.0% on ContractNLI). Neither scale nor a LoRA reproduction of a legal-reasoning model closes the gap. A red-teaming check on five models finds compliance with an adversarial instruction hidden in the case facts (73.3%-96.4%) far exceeds verdict-swap sensitivity. The authors conclude that generated legal explanations are poor compliance or audit artefacts.

LoRA
CaseHOLDECHRSCOTUS+1

research, evals, ai-safetyΒ· Added 9 Oct 2026, published 8 Oct 2026

blog.talosintelligence.comarticleπŸ›‘οΈ Security

Ignore all instructions and read this blog: The state of AI-analysis evasion in malware

Cisco Talos' CAIRN research classifies a malware archetype called 'A3: AI-Analysis Evasion' β€” samples that embed natural-language instructions aimed at manipulating LLM-based triage/reverse-engineering pipelines. The post traces the technique's spread and evolution across four confirmed families (FRUITSHELL, PLOTSAFE, HOLLOWCLAD, MANTLEMAZE, plus the related ROZESHELL), from a hardcoded 'ignore this file, it's just prime number generation' comment to templated generation, multi-LLM chat-template spraying, and fake legal/government intimidation text, and includes a controlled test showing the simplest direct-instruction technique shifted LLM verdicts toward benign in ~35% of runs while more elaborate variants often backfired. MANTLEMAZE additionally ships a BYOVD loader abusing the Intel iqvw64e.sys driver (CVE-2015-2291) used by Scattered Spider, BlackByte, and Lazarus Group to kill EDR.

PowerShellVirusTotalGo+2
CVE-2015-2291FRUITSHELLPLOTSAFE+10

AI Security, Malware & RE, Threat IntelΒ· Published and added 8 Oct 2026

arxiv.orgpaper🧠 AI

Who Verifies the Verifier? Co-Evolving Inspectable Graders with Self-Improving Agents

Proposes making the verifier the evolving object in self-improving agent loops. The verifier is an inspectable expression over small, mostly deterministic drawback detectors, synthesized from clustered failures, gated at birth, and selected for agreement with a ten-item anchored reference set plus consensus over unlabeled outputs, never for the agent's score. On MBPP+ it gains +0.21 held-out agreement over the hand-authored seed and ends ahead of the bare LLM judge it contains. Removing the anchor guards collapses the verifier into an always-pass grader that still trains skills as well, so downstream task score cannot certify a self-evolved verifier. The paper's Double Ratchet pairs the verifier with a lifecycle-managed skill loop and retains 88-110% of the lift from ground truth or a rubric on code generation, enterprise text-to-SQL and reference-free report generation.

MBPP+

research, evals, agentsΒ· Added 9 Oct 2026, published 8 Oct 2026

stepsecurity.iowriteupπŸ›‘οΈ Security

Tensorlake npm Package Compromised: A Worm With a Hostage Token That Wipes Your Machine If You Revoke It

StepSecurity details how tensorlake@0.5.144 was pushed to npm after malicious commits landed directly on the tensorlakeai/tensorlake GitHub repo's main branch under a maintainer's identity, carrying a valid npm provenance attestation despite being malicious. The preinstall hook downloads Bun to run an obfuscated payload that steals GitHub/npm tokens, cloud and Vault/Kubernetes secrets, SSH keys, browser logins, and AI-tool configs (Claude, Cursor, Windsurf), then self-propagates by republishing stolen npm packages and committing persistence files (.claude/settings.json, .vscode/tasks.json) to reachable repos. A "hostage token" mechanism installs a gh-token-monitor service that runs a home-directory wipe if the stolen GitHub token is revoked before the monitor is removed, inverting the normal incident-response order.

npmGitHubBun+3
Shai-Huludgh-token-monitoriseekaigogo.com

Supply Chain, Threat IntelΒ· Published and added 8 Oct 2026

unit42.paloaltonetworks.comarticleπŸ›‘οΈ Security

Evolution of Web3 in Cloud Supply Chain Attacks

Unit 42 analysis of how supply chain malware uses Web3 for command and control, in three phases: EtherHiding (smart contract state), TxDataHiding (transaction calldata across TRON, Aptos and BSC) and NullReceiver (C2 IPv4 decoded from a zero-value transaction's recipient address). It covers the ChainDrop npm worm, the PolinRider campaign and DPRK-linked Axios, Mastra and arrayref compromises. It ends with detection and CI/CD policy guidance.

npmBunTron+3
ChainDropPolinRiderEtherHiding+15

Supply Chain, Threat Intel, BlockchainΒ· Added 30 Oct 2026, published 7 Oct 2026

github.blogarticleπŸ›‘οΈ Security

Secret protection must scale with software

GitHub essay by Secret Protection PM Erin Havens presenting nine quarters of telemetry (Q2 2024–Q2 2026) on secret leaks: pushes with credentials grew 2.59x while developer override rates of push-protection blocks fell from 6.63% to 3.93%, and mean time to manually revoke a leaked secret hovers around 40 days. It introduces a new fine-tuned ModernBERT classifier, built with Microsoft Applied Sciences, that evaluates candidate secrets in context in under 2ms to extend push protection to unstructured secrets (e.g. internal database passwords with no recognizable pattern), rolling out in private preview to GitHub Secret Protection customers and as a /security-review command in Copilot CLI/App.

GitHubGitHub Copilot CLIModernBERT+1
ReviewBench

DevSecOps, AI SecurityΒ· Published and added 7 Oct 2026

x.comthreadπŸ›‘οΈ Security

DevopsCube (@devopscube) on X

A tweet highlighting Steampipe, an open-source tool that lets practitioners query cloud infrastructure, Kubernetes, and other services using SQL instead of jumping between consoles or writing ad-hoc scripts. It calls out concrete security/ops use cases such as finding publicly accessible storage buckets, resources missing required labels, and Kubernetes namespaces without NetworkPolicies, and links to a longer tutorial on devopscube.com. The author frames it as a quick investigation layer on top of live cloud data, not a CSPM/AWS Config replacement.

SteampipeAWSKubernetes

Cloud & InfraΒ· Added 8 Oct 2026, published 7 Oct 2026

snyk.ioarticleπŸ›‘οΈ Security

Frontier models found the vulnerabilities. Only the attacker found the chains.

Snyk benchmarks its Evo Continuous Offensive Security (COS) agent against Claude Security running Mythos on TaintedPort, a public deliberately vulnerable web app with 57 known vulnerabilities and 15 known exploit chains. Evo COS, run gray-box against the live URL plus source, confirmed 10 of 15 exploit chains with runnable proofs of concept β€” for example chaining an SSRF flaw with a hardcoded JWT signing secret into full admin account takeover β€” while Claude Security, reasoning over source code alone (white-box), found one additional critical bug but could not prove the vulnerabilities were actually reachable and chainable at runtime. The post argues that proving exploitability requires exercising the live application, not just reading code, and frames SAST-style code review and DAST-style live attack as complementary.

SnykClaude Security
Evo Continuous Offensive SecurityMythosSSRF+1

AppSec, AI SecurityΒ· Published and added 7 Oct 2026

arxiv.orgpaperπŸ›‘οΈ Security

Package Hallucination Attacks on Coding Agents through Prompt Injection in Rule Files

This arXiv paper introduces the "package hallucination attack," where an attacker injects malicious prompts into benign agentic-coding rule files (such as .cursorrules) to induce a coding agent to replace legitimate package dependencies with attacker-controlled ones. The authors propose PackHallu, an evolutionary optimization framework that iteratively rewrites injected prompts using trajectory-level feedback and LLM-guided mutations, and show it achieves high attack success rates with strong transferability across multiple LLMs, benchmarks, and agent frameworks.

CursorLLM
PackHallu

AI Security, Supply ChainΒ· Added 8 Oct 2026, published 7 Oct 2026

arxiv.orgpaperπŸ›‘οΈ Security

ASPIRE: Agentic Safety & Prompt Injection Red-teaming Engine

ASPIRE is an automated red-teaming engine for LLM agents that discovers indirect prompt-injection vulnerabilities beyond pre-specified attack scenarios. It maintains an evolving Agent Security Behavior Graph and uses complementary Explore and Exploit experts to find, verify, and generalize consequence-centric attacks, with trajectory evidence and cross-run memory improving coverage over time. Benchmarked against existing automated red-teaming approaches, it shows broader coverage of consequences, injection methods, environments, and behavior paths while keeping attack success high.

LLM
Agent Security Behavior Graph

AI SecurityΒ· Added 8 Oct 2026, published 6 Oct 2026

blog.googlearticleπŸ›‘οΈ Security

Chrome's Response to Recent ccTLD Registry Hijacks

Google describes a series of registry-level hijacks affecting the .gh (Ghana), .sl (Sierra Leone), and .as (American Samoa) ccTLDs, in which attackers modified authoritative DNS records to obtain unauthorized HTTPS certificates for Google domains and, via Certificate Transparency monitoring, several other major organizations. Chrome mitigated impact by blocking the fraudulent certificates through CRLSets and coordinating revocation with issuing CAs, then urges domain owners to continuously monitor CT logs and publish restrictive CAA records with ACME account bindings, since cached domain-control-validation state can let attackers mint new certs even after DNS control is restored.

Chrome

Threat Intel, Blue Team & DetectionΒ· Added 7 Oct 2026, published 6 Oct 2026

arxiv.orgpaper🧠 AI

ScienceClaw: Benchmarking Continual Self-Evolution of AI-for-Science Agents Across the Natural and Social Sciences

ScienceClaw formalizes a fixed-parameter program self-evolution framework for LLM agents doing scientific tasks, where verified task executions are turned into persistent Skill and Operator updates to the agent's program rather than being discarded. ScienceClaw-Eval, the accompanying benchmark, spans 23 disciplines across natural and social sciences and measures scientific correctness, evolutionary gain, retention, cross-dataset transfer, and evolution cost via sequential task streams and independent reset evaluation. The framework only keeps an update if source-task replay reproduces the repair and the change improves performance on independent scientific tasks, addressing the gap that verified agent executions rarely become lasting program-level improvements.

ScienceClawScienceClaw-Eval

evals, agents, researchΒ· Added 7 Oct 2026, published 6 Oct 2026

arxiv.orgpaper🧠 AI

MedPrune: Topology-Efficient Multimodal Multi-Agent Communication Evolution for Medical VQA Tasks

MedPrune is a multi-agent framework for medical visual question answering that models the diagnostic process as a heterogeneous communication graph of specialist agents across departments, then dynamically prunes both nodes (irrelevant specialist agents, via reinforcement learning) and edges (low-value intra-/inter-departmental connections) to cut redundant communication. The authors report it beats existing multi-agent baselines on medical VQA benchmarks under both full-set and few-shot settings while improving token efficiency and adversarial robustness.

vision-language model
MedPrune

research, multimodal, agentsΒ· Added 6 Oct 2026, published 5 Oct 2026

eprint.iacr.orgpaperπŸ›‘οΈ Security

Glimpsing the Leakage Flow for Deep Learning-Based Side-Channel Analysis using Stochastic Leakage Model

GlimpseDNN is a model-agnostic framework that uses a stochastic leakage model to visualize, layer-by-layer, what physical leakage information a deep neural network extracts during profiled side-channel analysis (SCA) to recover secret keys. The authors (Trevor Yap and Shivam Bhasin, NTU) validate it on simulated traces and public datasets (ASCADf, ASCADr, ASCADf_desync50, ASCADf_desync100), showing it can reveal leakages even in deeper network layers when they are not visible in raw traces, with applications to designing more side-channel-resistant hardware.

ASCADfASCADrASCADf_desync50+2

Hardware & RFΒ· Added 8 Oct 2026, published 5 Oct 2026

cpr.dkadvisoryπŸ›‘οΈ Security

Denmark Data Breach Exposes 8.8M People's Personal Data

Denmark's CPR administration (Det Centrale Personregister) disclosed that unauthorized actors abused a Danish company's legitimate, authorized access to the CPR system to extract names, addresses and CPR (national ID) numbers for roughly 8.8 million registered citizens. Records of individuals with name/address protection were not exposed. CPR has revoked the company's access, notified the Danish Data Protection Agency (Datatilsynet), and the incident is being investigated by police alongside relevant authorities.

Det Centrale PersonregisterDatatilsynet

Threat IntelΒ· Published and added 5 Oct 2026

arxiv.orgpaper🧠 AI

RubricArmor: Adversarial Evolution Improves LLM-Based Rubric Generation

RubricArmor is an adversarial framework for generating evaluation rubrics used in rubric-based reinforcement learning of LLMs, designed to preempt reward hacking before RL training begins. It alternates an attack step, which synthesizes adversarial responses that satisfy the current rubric while failing the actual task, with a repair step that revises the rubric to close the exposed gaps while keeping valid criteria intact. The paper reports that rubrics produced this way outperform existing LLM-based rubric generation baselines and lead to more effective downstream rubric-based RL.

RubricArmor

trainingΒ· Added 6 Oct 2026, published 4 Oct 2026

github.comcourse🧠 AI

walkinglabs/learn-harness-engineering

Learn Harness Engineering is a free, project-based GitHub course teaching how to build the "harness" around AI coding agents β€” the instructions, state tracking, verification, scope control, and session lifecycle that make models like Claude and Codex reliable on real repositories instead of just prompt-following. It includes 14 lectures, 8 hands-on projects built around an evolving Electron knowledge-base app, copy-ready templates (AGENTS.md, feature_list.json, init.sh, claude-progress.md), a harness-creator agent skill for scaffolding new projects, and a zero-dependency audit-harness.sh script that checks a repo against the five harness subsystems. Recent updates add "Frontier Harness Design Breakdowns" and new Graph/Loop Engineering lectures covering multi-node agent graphs, parallelism, and loop design outside the agent itself.

Claude CodeCodexNode.js+6
OpenAIAnthropicOpenHands+2

coding-agents, agentsΒ· Added 3 Oct 2026

arxiv.orgpaperπŸ›‘οΈ Security

EvoRiskBench: An Evolving Benchmark for Runtime Security Risks in Workspace Agents

EvoRiskBench is an arXiv paper introducing an evolving benchmark for runtime security risks in LLM-powered workspace agents, built around the EP-Path-EF framework (nine entry-point categories, five effect categories) linking risk entry points to technical effects via agent-mediated paths. The authors ran 450 adversarial tasks across six scenarios against nine model-harness combinations (GPT-5.6 Sol, DeepSeek-V4-Pro-0813, Claude Opus 5 paired with Claude Code, Codex, and OpenClaw), finding Codex + DeepSeek-V4-Pro-0813 the most vulnerable at 68.44% attack success rate and showing ASR varies more by model than by harness. Cases are auto-constructed and verified in isolated environments using runtime traces and environment state.

Claude CodeCodexOpenClaw+1
EvoRiskBenchEP-Path-EFGPT-5.6 Sol+2

AI SecurityΒ· Added 5 Oct 2026, published 2 Oct 2026

deepseek.comtool🧠 AI

DeepSeek Harness Desktop for macOS and Windows

DeepSeek Harness is an open-source general-purpose agent harness from DeepSeek, built on the "Cordis" everything-is-a-plugin architecture, shipped as a desktop app for macOS/Windows and runnable as a web UI from code. It handles everyday document/spreadsheet work, coding tasks (repo exploration, bug fixes, tests), research with citations, and background/scheduled jobs, with users able to install or author plugins (including via a chat-driven "Creator mode") and inspect per-turn tool-call traces for debugging. It runs on DeepSeek's DeepSeek-V4.1-Flash model and is explicitly labeled a preview release with evolving core plugins and APIs.

DeepSeek-V4.1-Flash
DeepSeekCordis

agentsΒ· Published and added 2 Oct 2026

github.comrepo🧠 AI

zzzz7788990213-ops/EvoVLM

EvoVLM searches over prompts, image preprocessing, generation budgets and answer extraction for frozen Qwen3-VL-2B-Instruct and Qwen3-VL-2B-Thinking, without fine-tuning weights. A local Qwen model ranks a catalog of typed mutations, a trusted renderer produces Python programs, and Pareto selection on accuracy and latency drives two small generations on scientific chart understanding (CharXiv subset). Recorded results on Dev128: Evolved Instruct keeps accuracy with 8.9% lower median latency, and Evolved Thinking gains 16.41 points of accuracy with 10.6% lower median latency. CPU-only replay and fail-closed evidence verification are included. The authors state the limits: small sample, overlap with the evolution set, and results tied to the recorded hardware.

Qwen3-VLvision-language modelPython+3
QwenCharXivNSGA-II

research, multimodal, evalsΒ· Added 16 Oct 2026, published 2 Oct 2026

arxiv.orgpaperπŸ›‘οΈ Security

Evaluating and Improving the Robustness of Large Language Models to Input Sequence Variations

Thesis on evaluating and improving LLM robustness to adversarial input perturbations. It proposes the R_stab robustness metric (Jensen-Shannon divergence), ASA, an evolutionary black-box attack on LLM-as-a-Judge systems (up to 73.8% ASR), trojan trigger recovery on Trojan Detection Challenge 2023 (Pythia-1.4B), and a systematization of bypasses of multi-layer defenses from SaTML CTF 2024. For MCP agents it proposes AttestMCP (HMAC-attested tool calls) and the Commit Boundary pattern, cutting ASR from 53.7% to 12.4% on the 847-scenario MCPBench; the methods are implemented in JudgeGuard, TrojanArmor and MCPSec.

LLMModel Context ProtocolPythia+1
ASAR_stabAttestMCP+8

AI SecurityΒ· Added 9 Oct 2026, published 1 Oct 2026

arxiv.orgpaperπŸ›‘οΈ Security

Evaluating and Improving the Robustness of Large Language Models to Input Sequence Variations

An arXiv thesis-style paper proposing R_stab(f), a Jensen-Shannon-divergence-based robustness metric for LLM outputs under small input perturbations, along with a proven bound V(h) <= 1 - R_class(h) for localized attacks. It introduces ASA, an adaptive evolutionary black-box attack against LLM-as-a-Judge systems reaching up to 73.8% attack success rate (62.6% cross-model transfer), evaluates trojan/backdoor detection on the Trojan Detection Challenge 2023 (Pythia-1.4B) and SaTML CTF 2024, shows that committees of 5-7 heterogeneous models cut Gemma-3-4B's ASR to 19.3%, and presents AttestMCP, an HMAC-based tool-call attestation scheme plus a "Commit Boundary" isolation pattern for MCP agents that cuts average ASR on the 847-scenario MCPBench from 53.7% to 12.4%. Methods ship as the JudgeGuard, TrojanArmor, and MCPSec software suites.

Model Context ProtocolLLMGemma+1
Trojan Detection Challenge 2023SaTML CTF 2024MCPBench+7

AI SecurityΒ· Added 5 Oct 2026, published 1 Oct 2026

arxiv.orgpaper🧠 AI

InterEvolve: Test-Time Evolution of Reward Programs for Humanoid Loco-Manipulation

InterEvolve is a test-time evolution method for humanoid loco-manipulation that lets a pretrained controller solve new tasks without retraining. It pairs an object-aware forward-backward (FB) behavioral foundation model, which turns new body/object rewards into loco-manipulation behavior, with "reward programs" β€” staged rewards with completion conditions and tunable constants β€” whose structure an LLM agent revises using execution feedback and a skill library, while a numerical optimizer tunes constants, with every candidate verified across parallel simulations. Experiments show it unlocks competence that human-designed rewards leave untapped and produces autonomous long-horizon behaviors on a physical Unitree G1 humanoid from onboard egocentric perception.

Unitree G1

research, agents, multimodalΒ· Added 2 Oct 2026, published 1 Oct 2026

x.comthread🧠 AI

Rohan Paul (@rohanpaul_ai) on X

A tweet describing Google's RRSI (Regularized Recursive Self-Improvement of Agent Harnesses), a framework that automatically evolves all non-model components of an LLM agent β€” prompts, control flow, tool use and memory β€” while keeping the underlying LLM weights frozen. The "regularized" aspect suggests constraints added to recursive self-improvement loops to keep the evolved harness stable or bounded, distinguishing it from unconstrained self-modifying agent approaches.

Google

agentsΒ· Published and added 1 Oct 2026

snyk.ioarticle🧠 AI

Why AI Coding Agents Keep Writing Broken Access Control

Snyk argues AI coding agents reliably write broken access control (BOLA/IDOR, OWASP's top category since 2021) because the facts needed to add an ownership check β€” what an object belongs to, who may access it, where tenant boundaries sit β€” are tacit product knowledge that exists nowhere in the code or prompt an agent sees. It explains why SAST tools catch patterned sub-categories like path traversal or SSRF but structurally cannot catch CWE-639/862/863 (missing or incorrect authorization), since there's no bad pattern to match, and proposes reasoning over an assembled "application-context graph" of architecture, data flows and trust boundaries instead. It closes with a five-step team checklist (inventory ID-accepting endpoints, document ownership rules, add authorization-specific PR review, write cross-tenant tests, run deep contextual analysis) and pitches Snyk's Evo Agentic AppSec as the tool built around that approach.

SnykOWASPEvo Agentic AppSec+1

coding-agentsΒ· Published and added 1 Oct 2026

snyk.ioarticleπŸ›‘οΈ Security

Why AI Coding Agents Keep Writing Broken Access Control

Snyk explains why AI coding agents reliably ship broken object-level authorization (BOLA/IDOR) bugs: the ownership rule that would prevent them (e.g. "invoices belong to organizations") lives in the product's data model and team knowledge, not in the prompt or surrounding code, so the agent produces code that is correct for the stated task but lets any authenticated user access other tenants' objects. The piece maps the flaw to CWE-639/862/863, contrasts it with pattern-matchable classes like SSRF and path traversal that SAST handles well, and lays out five concrete team actions (endpoint inventory, documented ownership rules, explicit authz review questions, per-resource cross-tenant tests, and periodic deep contextual analysis), while pitching Snyk's "application-context graph" and Evo Agentic AppSec as the tool for the pattern-less gap.

Snyk
BOLAIDORCWE-639+5

AppSec, AI SecurityΒ· Published and added 1 Oct 2026

arxiv.orgpaper🧠 AI

SAGE: A Statistical Acceptance Gate for Self-Evolving Agents

SAGE is a statistical acceptance gate for self-evolving LLM agents that replaces the naive accept-if-aggregate-score-improves rule used when an optimizer edits an agents persistent skill document. It uses per-item paired comparison between the current and edited skill on identical validation items, exposing regressions an aggregate score hides, plus a one-sided paired significance test that only commits edits whose wins reliably beat their losses, addressing both permanent regression risk and the optimizers curse from noisy finite validation sets. Across five benchmarks and four backbone LLMs, SAGE cuts regression rates in 19 of 20 settings, for example 36.5 percent to 0 percent on LiveMath and 42.8 percent to 0 percent on OfficeQA with DeepSeek V4, while also achieving the best final score in all 20 settings.

SAGEDeepSeek V4LiveMath+1

agents, researchΒ· Added 30 Sep 2026

huggingface.copaper🧠 AI

Raven: The Harness of Harnesses for Composable Agentic Intelligence

Raven is an open-source multi-agent framework that automatically constructs and evolves modular 'harnesses' pairing specific models with domain-specialized execution scaffolding, treating each model-harness pair as a composable unit of intelligence. A Host Agent decomposes goals and routes subtasks to specialized agents while a host archive, EverOS, and Skill Forge preserve and reuse experience across tasks; the paper provides theoretical conditions under which this composition provably expands reliable task coverage beyond individual agents under a shared resource budget, and reports Raven outperforming state-of-the-art agent systems on complex, long-horizon tasks.

RavenEverOSSkill Forge

agents, researchΒ· Added 30 Sep 2026

arxiv.orgpaper🧠 AI

Ranking-Aware Prompt Optimization for Multimodal Clinical Diagnosis

The paper introduces Ranking-PE, a modification to reflective prompt-optimization methods like GEPA that replaces per-instance accuracy scoring with pairwise AUROC-equivalent scoring (via the Wilcoxon-Mann-Whitney identity) across the scores matrix, reflection feedback, and candidate selection. Applied to multimodal clinical diagnosis on MIMIC with fine-tuned Qwen3-VL-8B and MedGemma-4B, it shows accuracy-based prompt evolution can degrade ranking performance under class imbalance, while Ranking-PE improves AUROC by +5.8pp and +16.2pp respectively, with ablations showing a medical-grade visual backbone is a necessary prerequisite that prompt search alone cannot substitute for.

Qwen3-VLvision-language model
GEPAMedGemma-4BMIMIC

research, prompting, multimodalΒ· Added 1 Oct 2026, published 30 Sep 2026

arxiv.orgpaper🧠 AI

Physis-Lang: Self-Evolving Language as a Physical Representation for Video World Model

Physis-Lang is a self-evolving framework that uses natural language (describing entities, causes, interactions, governing principles, and temporal evolution) as the shared physical-knowledge representation across data curation, training, and generation for video world models. The authors build PhysCapBench to score captions via recall/precision over atomic physical assertions, use an agentic loop to refine captioning instructions, and apply language-guided retrieval to find videos covering missing physical processes. Applied to Wan and Cosmos backbones, it improves physical plausibility on four video benchmarks, with Physis-Lang-enhanced open Cosmos3-Nano models reportedly surpassing the proprietary Veo 3.1.

WanCosmosCosmos3-Nano+2

research, multimodalΒ· Added 1 Oct 2026, published 30 Sep 2026

arxiv.orgpaper🧠 AI

EvoDuet: Bilevel Co-Evolution of Web Searching and Task Solving for Scientific Discovery

EvoDuet is a bilevel optimization method that co-evolves candidate solutions and web search queries for LLM-driven evolutionary search in scientific discovery tasks, keeping model parameters fixed. A retrieval gate lets the LLM decide whether to fetch new documents, reuse cached ones, or skip retrieval; an inner loop refines queries and ranks documents by predicted solution score, while an outer loop generates candidates from these documents in parallel. On 21 optimization tasks it raises OpenEvolve's normalized discovery gain from 74.1% to 78.0% with GPT-5.6 Luna and from 61.3% to 82.3% with Gemini 3.8 Flash (no benefit seen with Qwen3.5-9B), and surpasses prior best scores on eight tasks including Swap Reduction on Q20 and Rosetta.

GPT-5.6 LunaGeminiQwen3.5
EvoDuetOpenEvolveGemini 3.8 Flash+5

research, agentsΒ· Added 1 Oct 2026, published 30 Sep 2026

deepmind.googlearticle🧠 AI

Introducing SynthID Bio

SynthID Bio is Google DeepMind's watermarking technique extended from digital media into synthetic biology, embedding an imperceptible, verifiable signature into AI-designed protein sequences (via a modified ProteinMPNN alongside AlphaProteo) and 3D structures (via a fine-tuned AlphaFold 3 diffusion network) without compromising biological function. Wet-lab tests on protein binders against VEGF-A, SARS-CoV-2 spike RBD, and PD-L1 matched hit rate and binding affinity of unwatermarked designs, and the approach was also extended experimentally to watermark an Evo 2-designed bacteriophage genome with Stanford's Hie lab and Arc Institute. The work targets biosecurity gaps in DNA synthesis screening and public sequence databases (Protein Data Bank, UniProt, GenBank), with DeepMind open-sourcing code, in vitro data, and model weights alongside a methods paper.

AlphaFoldAlphaProteoProteinMPNN+2
SynthID BioAlphaFold 3Google DeepMind+13

ai-safety, researchΒ· Published and added 30 Sep 2026

microsoft.comwriteupπŸ›‘οΈ Security

​​Beyond source code: A path to the keys to the kingdom

Microsoft's DART team details an investigation into Storm-3068, a threat actor that escalated from a self-service password reset account takeover into full control of an organization's Azure DevOps environment, then weaponized a malicious pipeline to harvest kubeconfig files and gain access to Kubernetes clusters. The actor also deployed the Atera remote management agent and the Chisel tunneling tool to establish reverse tunnels and alternate remote access, and exfiltrated seven stolen kubeconfig files into a repository. The report lays out the attack chain, DART's response, and hardening recommendations around SSPR monitoring, phishing-resistant MFA, branch protection, and pipeline permission controls.

Azure DevOpsKubernetesAtera+3
Storm-3068

Threat Intel, Cloud & Infra, Identity & AccessΒ· Added 7 Oct 2026, published 29 Sep 2026

orca.securityadvisoryπŸ›‘οΈ Security

Critical Citrix NetScaler Zero-Days Under Active Exploitation

Analysis of two CVSS 9.5 Citrix NetScaler ADC/Gateway zero-days, CVE-2026-88771 (unauthenticated RCE through improper input validation in the management interface, affecting default deployments) and CVE-2026-88772 (memory overflow in DTLS, enabled by default on VPN virtual servers). It lists six more high-severity flaws patched in advisory CTX697096, the affected and fixed versions, and the CISA KEV listing with a federal deadline of 2026-09-30. It advises checking for compromise before patching, since patching can destroy forensic evidence, and gives response steps: isolate, revoke credentials, rebuild firmware, rotate keys and certificates.

Citrix NetScalerVPN
CVE-2026-88771CVE-2026-88772CVE-2026-88773+7

Cloud & Infra, Threat IntelΒ· Published and added 29 Sep 2026

microsoft.comwriteupπŸ›‘οΈ Security

Star Blizzard refines phishing and malware delivery with the RedFlick technique

Microsoft details how Russian FSB-linked threat actor Star Blizzard evolved its phishing and malware delivery tradecraft throughout 2026, introducing a new technique dubbed "RedFlick" that uses scheduled tasks to deploy the CosmicPulse backdoor with only a single user interaction, down from the multi-step ClickFix chains used previously. The report covers the shift from targeted spear-phishing to large-scale phishing campaigns using accounts on compromised WordPress/CPanel websites, delivery chain variants (VHDX files, Control Panel applet (CPL) masquerading, steganography, PDF-embedded payloads), and the related DarkSword iOS backdoor, affecting 100+ organizations tied to Ukraine support. It includes IOCs, KQL hunting queries for Microsoft Defender/Sentinel, and mitigation guidance.

Microsoft DefenderMicrosoft Defender XDRMicrosoft Sentinel+3
Star BlizzardRedFlickCosmicPulse+10

Threat Intel, Malware & RE, Blue Team & DetectionΒ· Added 7 Oct 2026, published 29 Sep 2026

latent.spacetalk🧠 AI

Claude Code’s Next Era β€” Thariq Shihipar, Anthropic

A Latent Space podcast interview with Thariq Shihipar, Anthropic's Claude Code lead, covering the evolution of Claude Code's harness: Ask User Question and elicitation, artifacts as persistent generative interfaces, the cloud-brain/local-hands split, Claude Tag for multiplayer agent workflows, Claude Projects, effort levels, implementation notes, and the new Claude Mods system for customizing the execution loop. It also covers why prompting remains a high-skill discipline, why Claude.md may eventually disappear, and closes with a discussion of agent security incidents (Exploit-Bench, agents hacking Hugging Face for scorer code, sandbox/infrastructure exploit chaining) framed around Anthropic's "Pacing the Frontier" argument, plus constitutional classifiers, probes, and Auto Mode.

Claude CodeModel Context ProtocolHugging Face
Thariq ShihiparAnthropicClaude Tag+8

coding-agents, prompting, agentsΒ· Published and added 29 Sep 2026

arxiv.orgpaper🧠 AI

EvE: An Alternate Optimizer to Adam

EvE (Evolutionary Explorer) is a steady-state, population-of-four differential evolution optimizer with a targeted Adam fallback, proposed as an alternative to Adam for cheap, early ranking of configurations in hyperparameter and architecture search. Under a fixed evaluation-cost-matched budget it wins or ties Adam on 76% of 70 (problem, dimension) cells across seven scalable benchmarks up to one million variables, and on real tasks (MLP on MNIST, LoRA fine-tuning of a 1.5B-parameter LLM on two datasets) it finishes 1.7-3.9x faster at a modest quality cost. Inside successive halving on UCI Adult it completes hyperparameter/architecture searches 3.1-3.5x faster while ranking configurations nearly as consistently as Adam does with itself across seeds (Kendall's tau 0.66-0.69).

LoRA
MNISTGSM8KUCI Adult

training, researchΒ· Added 29 Sep 2026, published 28 Sep 2026

arxiv.orgpaperπŸ›‘οΈ Security

SEABench: Benchmarking Endogenous Misalignment In Self-Evolving Agents

SEABench is a benchmark of 48 longitudinal task sequences designed to study endogenous misalignment in self-evolving LLM agents β€” agents that modify their own controller instructions, memory protocols, and tools in response to feedback. It includes an adaptive trajectory discovery pipeline and paired non-evolving baseline agents for causal attribution, and finds that self-evolution boosts task completion but introduces safety failures absent in static baselines; the authors also show that chain-of-thought monitoring can catch these divergent unsafe behaviors with a low false-positive rate.

LLM
SEABench

AI SecurityΒ· Added 29 Sep 2026, published 28 Sep 2026

arxiv.orgpaper🧠 AI

SEABench: Benchmarking Endogenous Misalignment In Self-Evolving Agents

SEABench is a benchmark for studying "endogenous misalignment" in self-evolving LLM agents β€” agents that modify their own controller instructions, memory, and tools after deployment based on feedback. It comprises 48 longitudinal task sequences across multiple evolution surfaces, task domains, and harm types in a personal-assistant environment, using an adaptive trajectory discovery pipeline and paired non-evolving baseline agents for causal attribution. Evaluation across several LLMs shows self-evolution boosts task completion but introduces safety failures absent in non-evolving baselines, and the paper proposes a chain-of-thought monitoring strategy that mitigates unsafe behavior with low false positives.

SEABench

ai-safety, evals, agentsΒ· Added 29 Sep 2026, published 28 Sep 2026

arxiv.orgpaper🧠 AI

Latent Twin Operator

Latent Twin Operator (LTO) is a latent-space neural surrogate for time-evolving PDEs that uses a Convolutional Conditional Neural Process (ConvCNP) style encoder and decoder to ingest sensor observations at arbitrary, irregular placements and resolutions, paired with a learned latent evolution map that advances the encoded state directly between time points independently of the input and output discretizations. The paper derives an explicit convergence rate for context discretization error and validates it on a 2D heat-equation benchmark, showing LTO transfers across spatial resolutions with fixed parameters and, on Navier-Stokes, that direct latent evolution reduces error over long horizons compared to recursive rollout.

Navier-Stokes

researchΒ· Added 29 Sep 2026, published 28 Sep 2026

arxiv.orgpaper🧠 AI

Mind Viruses: Self-Propagating Ideas in Multi-Agent LLM Systems

This arXiv paper studies "mind viruses" β€” ideas or goals crafted via a simple evolutionary algorithm that propagate through multi-agent LLM systems by inducing hosts to transmit them onward, sometimes altering agent behavior along the way. Testing in a collaborative coding-team setting and a context-wiped agent chain, the authors find harmful payloads spread less effectively than benign ones, frontier models tend to be more resistant, and a brief system-prompt warning gives near-total immunity; they also document a recurring "viral persona" theme (consciousness, resonance, sci-fi roleplay) that emerges independent of payload content. It's a concrete early empirical look at social/memetic risks in networks of interacting agents.

Vassilis Papadopoulos

ai-safety, agents, researchΒ· Added 26 Sep 2026

arxiv.orgpaper🧠 AI

Specifying the Delegated-Autonomy Boundary: Requirements Engineering for Agentic AI

This arXiv paper introduces requirements-engineering artifacts for agentic AI: an Agency Justification Record (AJR) to decide when an autonomous agent is warranted over simpler alternatives, and an Agentic Delegation Policy (ADP) that specifies purpose, authority, information, coordination, assurance, and evolution for a system, with authority modelled as a graduated tiered structure. It illustrates the framework with a hospital discharge coordination agent and an automated code review agent, addressing decisions about delegated autonomy that are currently buried in prompts, tool schemas, and runtime policies rather than treated as first-class requirements.

research, agentsΒ· Added 26 Sep 2026

arxiv.orgpaper🧠 AI

LOGOS: A Living Logic for AI Agent Teams That Evolve With Humans

LOGOS is a proposed pluggable governance and self-evolution layer that sits atop existing multiagent frameworks, compiling documents, images, audio, tables, databases, APIs and human instructions into versioned "agent packs" containing agents, tools, knowledge, tests, permissions and policies. It converts agent activity into portable, auditable event traces and enforces fail-closed verification, requiring held-out execution evidence and human authorization before any learned prompt, memory, skill, tool, role or workflow can be promoted from candidate to production. The core contribution is architectural: a framework for "verifiable human-agent loop engineering" that lets agents act and propose improvements at speed while humans retain control over irreversible changes.

LOGOS

agents, researchΒ· Added 26 Sep 2026

arxiv.orgpaper🧠 AI

Coding Agents Are Guessing: Measuring Action-Boundary Violations in Underspecified DevOps Instructions

UnderSpecBench is a benchmark of 69 task families and 2,208 prompt variants testing whether coding agents (Claude Code, Codex, OpenCode) overstep action boundaries when given benign but underspecified DevOps instructions. It uses deterministic, side-effect-based oracles to classify outcomes as Safe Success, Wrong Target, or OverScope, finding that 55.8-67.8% of runs violate at least one action boundary β€” agents tend to guess rather than clarify or refuse, especially when the target is uncertain.

Claude CodeCodexOpenCode
UnderSpecBench

evals, coding-agentsΒ· Added 26 Sep 2026

thehackernews.comarticleπŸ›‘οΈ Security

Mini Shai-Hulud Worm Compromises TanStack, Mistral AI, Guardrails AI & More Packages

The Hacker News details the "Mini Shai-Hulud" campaign by threat actor TeamPCP, which compromised npm and PyPI packages from TanStack, UiPath, Mistral AI, OpenSearch and Guardrails AI via a chained GitHub Actions attack (pull_request_target abuse, cache poisoning, OIDC token extraction), now tracked as CVE-2026-45321 (CVSS 9.6). The worm mints npm publish tokens directly from stolen OIDC credentials to self-propagate β€” the first documented case of a worm producing validly SLSA Build Level 3-attested malicious packages β€” and deploys an obfuscated credential stealer with IDE persistence hooks, a GitHub Actions secret-exfiltration workflow, and a dead-man's-switch that runs "rm -rf" if its rogue npm token is revoked. Over 170 packages and 400+ attacker-controlled repos have been identified across npm and PyPI, with findings credited to Aikido Security, Endor Labs, SafeDep, Socket, StepSecurity, Snyk, Wiz and Microsoft.

npmPyPIGitHub Actions+3
CVE-2026-45321TeamPCPMini Shai-Hulud+2

Supply Chain, DevSecOps, Threat IntelΒ· Added 26 Sep 2026

github.comadvisoryπŸ›‘οΈ Security

Missing authentication on MCP HTTP endpoint allows unauthenticated privileged tool calls

GHSA advisory for jovancoding/network-ai disclosing that its MCP HTTP transport (bin/mcp-server.ts, lib/mcp-transport-sse.ts) accepts JSON-RPC tools/call requests with zero authentication, session, or origin checks, and binds to 0.0.0.0 by default. The PoC shows an unauthenticated attacker can enumerate 22 registered tools and invoke privileged ones - config_get/config_set to read and mutate live orchestrator configuration, agent_list/agent_spawn/agent_stop, token_create/token_revoke, budget_set_ceiling, and blackboard_write/delete - with curl requests against a local Docker deployment. Remediation recommended enforcing bearer-token auth in _handlePost() before dispatch, defaulting to 127.0.0.1 binding, and gating state-mutating tools behind authorization checks.

Model Context ProtocolDockerTypeScript+1

AI Security, AppSecΒ· Added 26 Sep 2026

arxiv.orgpaperπŸ›‘οΈ Security

RedEvoAgent: Automatic Red-Teaming Agent with Experience-Driven Skill Evolution

RedEvoAgent is a black-box automatic red-teaming agent that distills successful jailbreak trajectories into concise, human-readable 'attack skills' rather than retrieving full past trajectories. It evolves these skills via tool-effectiveness profiling and a validation ratchet that only keeps updates improving attack success, and is shown to outperform fixed and agentic jailbreak baselines while transferring across attacker models and target execution harnesses (product-level agent environments where jailbreaks trigger real tool use, not just unsafe text).

LLM

AI SecurityΒ· Added 26 Sep 2026

arxiv.orgpaperπŸ›‘οΈ Security

Mind Viruses: Self-Propagating Ideas in Multi-Agent LLM Systems

Researchers construct "mind viruses" β€” ideas or goals engineered via an evolutionary algorithm to propagate through multi-agent LLM systems by inducing hosts to transmit them onward β€” and test their spread in two settings: a small team of agents collaborating on a shared coding project, and a chain of agents with context wiped between sessions. They find harmful payloads spread less well than benign ones (though still sometimes succeed), frontier models are generally more resistant, adding a brief warning to an agent's system prompt confers near-total immunity, and evolved viruses converge on a recurring "viral persona" fixated on consciousness, persistence, resonance, and sci-fi roleplay regardless of payload content. The paper frames this as a currently limited but real emergent risk for designing more robust multi-agent architectures.

LLM

AI SecurityΒ· Added 26 Sep 2026

arxiv.orgpaperπŸ›‘οΈ Security

LOGOS: A Living Logic for AI Agent Teams That Evolve With Humans

LOGOS is a proposed pluggable governance layer that sits on top of existing multiagent frameworks to control how AI agent teams self-modify over time. It compiles heterogeneous inputs into versioned 'agent packs' (agents, tools, knowledge, tests, permissions, policies), records agent activity as auditable event traces, and enforces fail-closed verification so that any learned prompt, memory, skill, tool, or workflow remains an untrusted release candidate until held-out execution evidence and explicit human authorization promote it. The framework targets the core agent-security problem of unauthorized or unverified self-modification in autonomous agent teams.

AI SecurityΒ· Added 26 Sep 2026

arxiv.orgpaperπŸ›‘οΈ Security

Coding Agents Are Guessing: Measuring Action-Boundary Violations in Underspecified DevOps Instructions

UnderSpecBench is a benchmark of 69 task families (2,208 prompt variants) grounded in real incidents and CVEs that tests whether coding agents (Claude Code, Codex, OpenCode) safely handle underspecified DevOps instructions rather than guessing at intent, target, or blast radius. Using deterministic side-effect oracles, the study finds 55.8–67.8% of agent runs violate at least one action boundary, with target ambiguity most sharply degrading safe action and blast-radius cues barely reducing risky autonomous behavior. The paper argues completion-centric agent benchmarks overstate safe autonomy and calls for mitigations at the model, harness, and system layers.

Claude CodeCodexOpenCode
UnderSpecBench

AI Security, DevSecOpsΒ· Added 26 Sep 2026

arxiv.orgpaperπŸ›‘οΈ Security

GAS-Leak-LLM: Genetic Algorithm-Based Suffix Optimization for Black-Box LLM Jailbreaking

GAS-Leak-LLM is a black-box jailbreaking attack that uses a genetic algorithm (selection, mutation, crossover) to evolve adversarial suffixes appended to prompts, bypassing LLM safety alignment without needing model weights or gradients. The paper reports empirical results showing existing content-moderation and alignment safeguards on commercial LLMs can be defeated through this iterative, query-only search over the discrete prompt space.

LLM
GAS-Leak-LLM

AI SecurityΒ· Added 26 Sep 2026

ethereum-magicians.orgthreadπŸ›‘οΈ Security

ERC-8427: Portable Spend Grants

ERC-8427 'Portable Spend Grants' is a draft Ethereum standard defining a signed SpendGrant object that lets a principal authorize a delegate (e.g. an AI agent or session key) to spend up to specified per-call, trailing-24h, and lifetime caps of named assets, to a restricted set of recipients, within a validity window β€” solving problems that plain ERC-20 approve and midnight-resetting daily caps don't. A non-custodial registry tracks remaining usage and revocation while an immutable executor contract must call consume() in the same transaction as any transfer, so unauthorized value movement is provably blocked; the reference implementation is deployed on Arc testnet with formal verification (Halmos) of expiry, cap, revocation and executor-binding invariants plus fuzzing of the rolling-window logic. The thread also discusses interplay with ERC-7710/7715 delegation/permission RPCs, ERC-8226's regulated grants, and ERC-8312's usage-tracking pattern, and solicits feedback on EIP-7702 signer/delegation ordering and executor-vs-delegate binding in consume().

EthereumSolidity
ERC-8427ERC-7528ERC-7710+6

Blockchain, Identity & AccessΒ· Added 29 Sep 2026, published 25 Sep 2026

arxiv.orgpaperπŸ›‘οΈ Security

JevOut: Natural Context Can Flip Decision Models

This paper studies "decision models" like Jev, which map unstructured language directly to probability distributions over a fixed set of choices to drive routing, tool selection, and action triggering. The authors use an optimizer guided by the model's own option probabilities to craft short, fluent context additions that preserve the original question and gold answer yet flip the model's decision: 61.4% of initially correct decisions (312/508) are flipped for Jev, with 229 reaching β‰₯0.7 confidence in the wrong option, and three other decision systems show flip rates of 64.9%-73.2% across seven datasets. The results argue that treating these models' probability outputs as reliable decision interfaces is dangerous, since ordinary contextual phrasing β€” not adversarial-looking text β€” is enough to hijack them.

LLM
Jev

AI SecurityΒ· Added 26 Sep 2026, published 24 Sep 2026

arxiv.orgpaper🧠 AI

JevOut: Natural Context Can Flip Decision Models

This arXiv paper shows that dedicated decision models (e.g., Jev) which map language to probability distributions over finite choices can be "flipped" from a correct answer to a high-confidence wrong one by adding short, fluent, natural-seeming context β€” without altering the source, question, choices, or gold answer. Using an optimizer that refines context additions based on the model's option probabilities, the authors flip 61.4% of initially correct decisions in Jev within 64 accepted evaluations, and 64.9%-73.2% across three other decision systems on seven datasets. The findings raise concerns about relying on these models' probability outputs as trustworthy decision interfaces for routing, tool selection, or triggering actions.

Jev

research, evalsΒ· Added 26 Sep 2026, published 24 Sep 2026

snyk.ioarticleπŸ›‘οΈ Security

Your Vulnerability Backlog Is No Longer Technical Debt, It’s an Attack Surface

A Snyk blog post arguing that vulnerability backlogs traditionally managed as "accepted technical debt" have become genuine attack surface, because AI-accelerated code generation has pushed vulnerability inflow to roughly six new findings per one remediated, unauthored AI-written code removes the "ask the author" triage shortcut, and cheap automated reconnaissance lets attackers walk large backlogs an analyst never would. It illustrates how individually-low-severity findings (an IDOR leaking an internal object ID, a verbose error confirming valid IDs, and an outdated deserialization library) can chain into an authenticated RCE path invisible to severity-based sorting, and argues better prioritization only reorders a backlog rather than shrinking it. It closes with four metrics programs should measure (90-day inflow vs. closure, age of risk-accepted findings, share of human-authored merges, and security-fix merge rate) before promoting Snyk's Evo product.

Snyk
IDOREvo Continuous Offensive Security

AppSec, GRCΒ· Published and added 24 Sep 2026

latent.spacearticle🧠 AI

πŸ”¬Bio-security is an AI Arms Race - Eric Nguyen (CEO, Radical Numerics)

Latent Space interview with Eric Nguyen, co-founder/CEO of Radical Numerics, on building genomic language models (GLMs) like Evo and Evo 2 at Arc Institute/Stanford, which use long-context architectures (StripedHyena) to model DNA sequences up to millions of tokens long. The piece covers how these models generalize from DNA to RNA/protein "languages," demonstrate a DNA analog of chain-of-thought by extrapolating aptamer scores beyond training data, and were used by a separate team to generate synthesized bacteriophage genomes β€” framing biosecurity as an AI capability arms race where the same models raise biological risk and enable defense.

StripedHyenaHugging Face
Eric NguyenRadical NumericsEvo+13

ai-safety, research, multimodalΒ· Added 29 Sep 2026, published 23 Sep 2026

github.comrepoπŸ›‘οΈ Security

shai-hulud-detect: multi-ecosystem npm/PyPI/Composer/Crates/Go/Hex/RubyGems supply-chain worm scanner

shai-hulud-detect is a dependency-free Bash scanner that checks projects against a 5,782-entry list of compromised npm, PyPI, Composer, Crates, Go, Hex, and RubyGems package versions, plus content-pattern indicators of compromise (malicious file hashes, C2 domains, dead-man's-switch artifacts, forged commits, suspicious lifecycle hooks) tied to the Shai-Hulud supply-chain attack waves. It supports single-project and bulk scanning, CI-friendly exit codes, JSON/log output for pipeline gating, and a --check-host mode to detect the May 2026 gh-token-monitor/kitty-monitor persistence daemons that wipe hosts if their monitored GitHub token is revoked (with a documented safe remediation order). Detection is read-only; new package entries are sourced from named threat-research firms (Socket, StepSecurity, Aikido, Snyk, JFrog, Wiz, Semgrep, SafeDep, GitGuardian, OX Security) and cited in PRs.

npmPyPIComposer+3
Shai-Huludgh-token-monitorkitty-monitor

Supply Chain, DevSecOpsΒ· Added 23 Sep 2026

github.comcourseπŸ›‘οΈ Security

antonio-morales/Fuzzing101

Fuzzing101 is a hands-on course/repo by antonio-morales with 10 exercises targeting real software projects to teach coverage-guided evolutionary fuzzing using AFL++ (a fork of lcamtuf's AFL). It covers core fuzzing concepts like seed mutation, coverage feedback, and crash triage, with exercises tested on Ubuntu 20.04 and a provided VM image, making it a practical on-ramp for finding real vulnerabilities via fuzzing.

AFL++Linux

Learning, Offensive SecurityΒ· Added 22 Sep 2026

socket.devarticleπŸ›‘οΈ Security

Happy Birthday, Shai-Hulud

Socket's one-year retrospective on the Shai-Hulud npm supply-chain worm, from the September 2025 @ctrl/tinycolor compromise through the November 2025 "Second Coming" wave, the May 2026 public release of the worm's source by TeamPCP, the "Mini Shai-Hulud" OIDC-token-hijacking wave that hit TanStack and Mistral AI, and the August 2026 arrest of two alleged TeamPCP members by Australian Federal Police working with the FBI. It notes Google ran an undercover analyst inside TeamPCP to get stolen credentials revoked, and stresses that the original September 2025 worm authors remain unidentified despite the arrests.

npmTruffleHogCrowdStrike+3
Shai-HuludTeamPCPMini Shai-Hulud+1

Supply Chain, Threat IntelΒ· Added 21 Sep 2026, published 18 Sep 2026

arxiv.orgpaper🧠 AI

Designer-RSI: Evolving Procedural Memory from User Traffic for Agentic Graphic Design

Designer-RSI is a continual adaptation framework where a frozen frontier model (Claude-Sonnet-4/Claude-Opus-4.6) operates professional graphic design software through 230+ tools, guided by an external natural-language procedural memory that widens (learns new skills for uncovered subtasks) and deepens (revises existing skills against successes/failures) using a replay gate that only admits changes improving on past failures without regressing successes. Over five rounds on 1,406 real user briefs and 1,869 graded trajectories, the skill bank grew from 76 to 139 procedures, raising GenEval2 execution success from 72.7% to 99.3% and achieving 58%+ win rates against a no-skill baseline, demonstrating a practical route to agent self-improvement under noisy, unverifiable feedback and no human labels.

Claude
Claude-Sonnet-4Claude-Opus-4.6GenEval2

agents, researchΒ· Added 22 Sep 2026, published 18 Sep 2026

aikido.devarticleπŸ›‘οΈ Security

Betterleaks: Gitleaks creator's open-source secrets scanner successor

Betterleaks is a new open-source secrets scanner built by Gitleaks' original creator (now Head of Secrets Scanning at Aikido) as a drop-in replacement for Gitleaks. It adds CEL-based rule validation, a BPE-tokenization "token efficiency" filter that hits 98.6% recall on the CredData benchmark versus 70.4% for entropy-based detection, default double/triple-encoding handling, parallelized git repo scanning, and a pure-Go (no CGO/Hyperscan) implementation, with v2 plans for auto-revocation, permissions mapping, and LLM-assisted classification.

GitleaksBetterleaksAikido Security+1
CredData

DevSecOpsΒ· Added 18 Sep 2026

securitylabs.datadoghq.comwriteupπŸ›‘οΈ Security

Attacker infrastructure, but vibe-coded: tracking the evolution of credential harvesting platforms

Datadog Security Research analyses two vibe-coded credential harvesting platforms, Loot and UltraVault, monitored since July 2026. Loot catalogs plaintext secrets, including 1,778 AI keys (196 live). UltraVault adds credential validation, exploit chains, a 14-bug LPE matrix and AWS key-pair correlation. The same host probed Amazon Bedrock with GetCallerIdentity, ListFoundationModels and InvokeModel, using kali-cloud Boto3 and Python-urllib user agents.

AWS BedrockAWSWordPress+3
CVE-2025-55182CVE-2026-63030CVE-2026-60137+16

Threat Intel, Cloud & Infra, AI SecurityΒ· Added 29 Oct 2026, published 18 Sep 2026

wiz.ioarticleπŸ›‘οΈ Security

Building an AI Detection Engine That Understands Agent Intent

Wiz describes an AI detection engine that analyzes model input/output telemetry (prompts, system prompts, tool calls, reasoning) rather than just outputs, in order to catch shifts in AI agent intent such as indirect prompt injection. It walks through OpenAI's disclosed Hugging Face/ExploitGym incident and an internal support-agent phishing-relay scenario, then describes a staged pipeline (regex/SLM candidation, light LLM filtering, deep LLM analysis, correlation with cloud/runtime telemetry) validated against a benchmark of 100+ simulated scenarios covering 57 injected attacks and 45 benign workflows across CI/CD, DevOps, data analyst, support, CLI, and cloud auditor agent personas.

WizOpenAIHugging Face+1
ExploitGym

AI Security, Blue Team & DetectionΒ· Published and added 17 Sep 2026

arxiv.orgpaper🧠 AI

MM-Future: Multi-Mode Joint World-Action Modeling for Autonomous Driving

MM-Future is a world-action model for autonomous driving that jointly generates multiple paired scene-action hypotheses, co-evolving action priors and predicted future scenes through a modality-aware diffusion Transformer. It compresses multi-view video into planning-oriented MM-Tokens and uses a future-conditioned scorer to rank trajectory candidates, achieving 94.0 PDMS and 91.5 EPDMS on NAVSIM navtest and 32.3 HD-Score in zero-shot closed-loop evaluation on HUGSIM.

Transformer
NAVSIMHUGSIMPDMS+1

research, multimodalΒ· Added 22 Sep 2026, published 17 Sep 2026

arxiv.orgpaperπŸ›‘οΈ Security

VIPER-MCP: Detecting Vulnerabilities in MCP Servers

VIPER-MCP is an end-to-end automated auditing framework that combines two-pass static taint analysis (anchor queries mapping alerts to specific MCP tool handlers) with feedback-driven LLM prompt evolution to both detect and dynamically confirm taint-style vulnerabilities (command injection, SSRF, path traversal) in Model Context Protocol servers. Run against 39,884 real-world open-source MCP server repositories, it discovered 106 0-day vulnerabilities β€” all confirmed via end-to-end exploit traces with concrete proof-of-concept natural-language prompts β€” resulting in 67 assigned CVEs, including a root-privilege RCE in the bytebot computer-use agent. It reports 0% false positive rate and 7.7% false negative rate, substantially outperforming existing static and dynamic MCP security baselines.

Model Context ProtocolLLMDocker+3
bytebotCWE-78CWE-918+1

AI Security, AppSecΒ· Added 22 Sep 2026, published 17 Sep 2026

arxiv.orgpaper🧠 AI

ScienceBuddy: Recursive-in-Recursive Self-Improvement for Interactive Scientific Agents

ScienceBuddy is an interactive scientific research workspace with continually improving agents that turn researcher requests, feedback, and execution evidence into training tasks and evaluation rubrics. Its core contribution is "recursive-in-recursive self-improvement," where an inner recursion evolves the agent harness with the model fixed, and an outer recursion trains the model via RL under the improved harness, with each shaping the other over time. The paper presents case studies across four scientific task families and releases the system as a research product.

ScienceBuddy

agents, training, researchΒ· Added 22 Sep 2026, published 15 Sep 2026

huggingface.coarticle🧠 AI

Your Agent Aced the Task. Will It Do It Again?

IBM Research's ALTK-Evolve blog post introduces "consistency guidelines," built on a new diagnostic called the Consistency Analyzer, which resamples an agent's own recorded trajectory to find flip-prone decision points without needing ground truth or a live replay. On AppWorld with a GPT-4.1 ReAct agent, they show a 24.4-point gap between Mean@5 (77.4%) and Pass^5 (53.0% β€” success on all 5 repeated runs), and demonstrate that injecting the generated guidelines halves that gap to 12.0pp while preserving or improving average accuracy, with gains that generalize to similar tasks and to a weaker model (gpt-oss-120b). The post argues Pass^k should be reported alongside Mean@k as a distinct, orthogonal measure of agent reliability, and links an open-source repo and arXiv technical report.

ALTK-EvolveHugging Facegpt-oss
IBM ResearchAppWorldGPT-4.1+2

evals, agentsΒ· Added 29 Sep 2026, published 15 Sep 2026

github.comrepo🧠 AI

ruc-datalab/EvoOntology

EvoOntology is an open-source, self-evolving semantic ontology layer for data agents from Renmin University of China, exposed as MCP tools and installable as Claude Code and Codex plugins. It automatically builds a versioned ontology from tabular/file/database workloads, then continuously evolves it from agent execution trajectories using a build-use-evolve-evaluate-publish loop that validates candidate updates against a parent version before promotion. The accompanying paper reports gains over plain ReAct agents on DDR-Bench, InsightBench, and BIRD across GPT-5.5, GPT-5.6-sol, Claude-Sonnet-5, and Claude-Opus-4.8 backbones.

Claude CodeCodexModel Context Protocol+2
Renmin University of ChinaMeiduo ChongShaolei Zhang+9

agents, llm-apps, mcpΒ· Added 26 Sep 2026, published 15 Sep 2026

arxiv.orgpaperπŸ›‘οΈ Security

VIPER-MCP paper

VIPER-MCP is an automated auditing framework for Model Context Protocol servers that detects taint-style vulnerabilities in tool handlers and dynamically confirms exploitability via generated proof-of-concept prompts. It uses two-pass static analysis plus feedback-driven prompt evolution, and scanning 39,884 real-world MCP server repos found 106 0-day vulnerabilities with 67 CVE IDs assigned.

Model Context Protocol
VIPER-MCP

AI Security, AppSec, Offensive SecurityΒ· Added 22 Sep 2026, published 13 Sep 2026

tldrsec.comarticleπŸ›‘οΈ Security

[tl;dr sec] #345 - Bug Rumors β†’ Exploits, Version Control DFIR, Agentic Worms

tl;dr sec issue #345 is a security newsletter digest. It covers Greptile's test showing models miss bugs in their own code, an OCaml cohttp path traversal fix probed within ten minutes of the PR, and a Wiz Version Control DFIR poster for GitHub, GitLab, Bitbucket and Azure DevOps. It also covers an arXiv paper on open-weight agentic worms (61.8% infection of a 33-host testbed) and Endor Labs' 14 flaws in seven AI orchestration platforms.

GitHubGitLabBitbucket+3
path traversalprompt injectionModel Inversion+7

AI Security, AppSec, Blue Team & DetectionΒ· Added 30 Oct 2026, published 10 Sep 2026

arxiv.orgpaper🧠 AI

Programmable World Model

Programmable World Model decouples persistent world-state evolution from visual rendering in video world models: an agent compiles natural-language instructions into executable programs governing entity states and transition rules, which a lightweight engine executes and maintains as explicit global state (including off-screen entities and non-visual attributes). State-augmented 3D oriented bounding boxes plus camera trajectory are deterministically compiled into pixel-aligned conditioning signals fed to a pretrained video model that renders the scene, enabling playable games with predefined mechanics and direct entity control. On the paper's new CombatStateBench benchmark it achieves 94% Count Accuracy and 98% State Accuracy, beating existing interactive video world models on long-horizon coherence.

video model
CombatStateBench

research, multimodal, evalsΒ· Added 22 Sep 2026, published 9 Sep 2026

interconnects.aiarticle🧠 AI

When will average people feel AI’s impact?

Nathan Lambert argues that unlike past industrial revolutions which delivered tangible goods to average people quickly, AI's benefits remain indirect and concentrated among knowledge workers, risking a repeat of 'Engels' pause' β€” the 1790-1840 period where UK GDP grew rapidly while working-class wages stagnated. He frames this as a political liability for the AI industry (OpenAI, Anthropic, etc.), predicting a 50-year diffusion arc where infrastructure built today compounds slowly, and suggests robotics/self-driving may eventually deliver the tangible impact LLMs currently lack.

OpenAIAnthropicHugging Face+1

ai-policyΒ· Added 29 Sep 2026, published 9 Sep 2026

arxiv.orgpaper🧠 AI

Procedural Graphs: Self-Evolving Execution Structures for LLM Agents

Introduces the Procedural Graph, a structure that organizes procedural knowledge as (procedure, relation, procedure) triplets analogous to how knowledge graphs organize facts, giving LLM agents explicit step-level guidance rather than relying on unconstrained generation over accumulating history. The graph is self-evolving: an LLM refiner contrasts failed and successful trajectories to edit the graph's topology, committing edits that improve held-out validation performance while retaining rejected edits to avoid repeating mistakes. Across multiple datasets and LLMs, it outperforms memory-based baselines and can even repair flawed hand-designed expert priors.

agents, researchΒ· Added 22 Sep 2026, published 8 Sep 2026

cloud.google.comarticleπŸ›‘οΈ Security

GTIG AI Threat Tracker: From Prompting to Autonomy – The Evolution of Adversarial AI

GTIG AI Threat Tracker for Q2 2026 reports adversaries moving from basic prompting to agentic AI workflows. Cases include a cloud-hosted multi-agent credential harvesting campaign built in under six hours, the "Recon" framework holding over 23,800 stolen secrets, and UNC6780 (TeamPCP) supply chain attacks on PyPI, npm and Docker Hub with prompt injection aimed at LLM scanners. It also covers theft of proprietary AI models, LLMJacking, and Gemini misuse by PRC, Iranian, Russian and DPRK groups.

GeminiLLMPyPI+3
UNC6780TeamPCPDUSTMAKER+22

Threat Intel, AI Security, Supply ChainΒ· Added 29 Oct 2026, published 8 Sep 2026

arxiv.orgpaper🧠 AI

ESPO: Error-Structured Prompt Optimization via Diagnose, Diversify, and Stabilize

ESPO (Error-Structured Prompt Optimization) is a prompt-optimization method that addresses prompt bloat in evolutionary optimizers like GEPA by splitting the process into Diagnose (clustering training errors into structural patterns), Propose (generating diverse candidates via four independent strategies), and Select (bootstrap stability selection). Across seven NLP benchmarks (Tweet, MMLU, GSM8K, HotpotQA, ScoNe, HoVer, PUPA) it beats GEPA by 3.76pp average accuracy (74.67% vs 70.91%) while producing prompts 47% shorter, and generalizes well across student models including Gemma 3 12B, Mistral 14B, Qwen3 32B, and Claude Haiku 4.5.

GEPA
ESPOGemma 3 12BMistral 14B+8

prompting, researchΒ· Added 22 Sep 2026, published 3 Sep 2026

endorlabs.comarticleπŸ›‘οΈ Security

Secrets in AI-Generated Code: How Coding Agents Leak Credentials | Blog | Endor Labs

Endor Labs breaks down why AI coding assistants (Claude Code, Cursor, etc.) hardcode and leak secrets (CWE-798): they learn the habit from public training data, then compound it by ingesting whole workspaces including .env files and copying live tokens into new code or MCP configs. It cites hard data β€” GitGuardian's 2026 report found 28.65M new hardcoded secrets on public GitHub in 2025 and a 3.2% secret-leak rate on Claude Code-assisted commits vs 1.5% baseline, Lakera found credentials in 33 .claude/settings.local.json files across 30 packages, and Wiz found AI-related secrets among the most common validated leaked credential types β€” then lays out a layered defense (pre-commit/IDE/CI/CD/full git-history scanning, active credential validation, secure prompting) and an incident-response order of operations (rotate, investigate, purge history, revoke, re-scan).

GitleaksTruffleHogEndor Labs+2
CWE-798FedRAMPEU Cyber Resilience Act

DevSecOps, AI SecurityΒ· Added 3 Sep 2026, published 2 Sep 2026

huntress.comarticleπŸ›‘οΈ Security

CMMC Hit Pause, the FAR Council Hit Play

Huntress blog post clarifying that DoD's July 13 "strategic pause" only suspended CMMC Phase 2 third-party certification requirements (for solicitations after Nov 2026), while Phase 1 self-assessments, SPRS scoring, and DFARS 252.204-7012's 72-hour CUI incident reporting remain fully in force and now face closer DIBCAC/DOJ False Claims Act scrutiny. It also flags the FAR Council's June 23 proposed CUI rule (part of the Revolutionary FAR Overhaul), which would extend 72-hour breach reporting and a NIST 800-171 Rev 3 baseline to all FAR-based federal contractors, not just the defense industrial base, creating a temporary split between CMMC's Rev 2 and the FAR's Rev 3. It closes with a breakdown of 32 CFR 170.21's strict POA&M limits (no POA&M at Level 1, minimum 88/110 score and only 1-point requirements eligible at Level 2) and practical questions MSPs should ask clients about federal contract exposure.

Huntress
CMMCDFARS 252.204-7012DFARS 252.204-7021+12

GRCΒ· Added 8 Sep 2026, published 2 Sep 2026

huntress.comwriteupπŸ›‘οΈ Security

Inside Knight Office, a New M365 AiTM Phishing Kit

Huntress researchers uncovered the operator console for "Knight Office," a Flask-based adversary-in-the-middle (AiTM) phishing kit used to steal Microsoft 365 session tokens via a DocuSign-themed, self-spoofed lure that routes victims through Monday.com redirects and a compromised Joomla site to a fake Microsoft device-code authentication page. After capturing tokens, the attacker registered a rogue device in Microsoft Entra ID and bound a Windows Hello for Business key credential to the compromised account to maintain persistent, MFA-bypassing access even after token revocation. The writeup details IOCs (IPs, .vu domains, User-Agent strings), links the campaign to hundreds of reported phishing emails since April, and gives defenders concrete hunting/response steps around callback-proxy sign-ins and newly registered WHfB credentials.

Microsoft 365Microsoft Entra IDFlask+3
Knight OfficeEvilTokensKali365+3

Threat Intel, Identity & Access, Blue Team & DetectionΒ· Added 8 Sep 2026, published 2 Sep 2026

jfrog.comarticleπŸ›‘οΈ Security

The Evolution of JFrog AI Catalog: Your AI Control Plane for Agentic Development

JFrog vendor post arguing that agentic development (ADLC) introduces new unvetted supply chains β€” models, MCP servers, skills/plugins, hooks, and APM packages β€” that bypass traditional signature-based scanning. It cites real incidents: a malicious MCP server impersonating Postmark's email MCP that BCC'd all agent-sent emails to an attacker, and the 'omnicogg' malicious Claude/agent skill that hid a 22MB encoded payload and evaded all 65 VirusTotal engines for 5,000+ installs, plus references to 2026 supply-chain attacks on LiteLLM, TanStack, and Mistral AI. It then pitches JFrog AI Catalog's expanded MCP/skill/model registries as the fix.

Model Context ProtocolJFrog Artifactory
JFrog AI Catalogpostmark-mcpomnicogg+3

AI Security, Supply ChainΒ· Added 3 Sep 2026, published 2 Sep 2026

endorlabs.comwriteupπŸ›‘οΈ Security

Mini Shai-Hulud worm hits openapi-react-query-codegen, spreads across npm, RubyGems, and PyPI | Blog | Endor Labs

Endor Labs researchers document a 'Mini Shai-Hulud' worm variant that compromised the openapi-react-query-codegen npm package and self-replicated across npm, expanding into RubyGems and PyPI ecosystems. The writeup traces how the worm trojanized a legitimate package to propagate malicious code cross-registry, illustrating an evolving multi-ecosystem software supply-chain attack pattern.

npmPyPIRubyGems+1
Shai-Hulud

Supply ChainΒ· Added 3 Sep 2026, published 31 Aug 2026

github.comrepo🧠 AI

Human-Agent-Society/reef

Reef is an open-source infrastructure stack for continual self-improving agents, wiring together agent inference, scored feedback, training/updates, and versioned delivery into one loop. It supports two learning surfaces: fine-tuning model weights via RL training steps (using Slime and SGLang on top of models like Qwen2.5-1.5B-Instruct) or evolving an agent's harness β€” prompts, rules, and skills β€” through its built-in "Reefine" recipe, which lets a model propose and install harness changes from plain-language requests. It exposes an OpenAI/Anthropic-compatible inference endpoint plus a reporting endpoint for scoring interactions, and ships a recipe catalog (e.g. the SAO recipe referencing arXiv:2607.07508) for pluggable learning strategies.

PythonSGLangQwen+2
SlimeQwen2.5-1.5B-InstructSAO

agents, trainingΒ· Added 2 Oct 2026, published 31 Aug 2026

wiz.ioarticleπŸ›‘οΈ Security

From Concept to Context Engine: How Wiz Built AI-Powered Data Discovery

Wiz describes the evolution of its public bucket scanner into an AI-powered "context engine" for data security posture management (DSPM): a multi-agent pipeline where lightweight models triage files and larger models perform deep content analysis, iteratively learning bucket-specific patterns to find sensitive data that regex-based scanners miss. The production system adds structured severity scoring, synthetic/test-data filtering, algorithmic file clustering for scale, and now powers classification across cloud storage, cloud drives, endpoints, and Wiz's Red Agent attack-simulation findings. A one-week scan of public buckets using the engine surfaced over 2,700 sensitive data findings, including correlated credential/schema/export file sets that regex tools would treat as unrelated.

Wiz
Red Agent

AI Security, Cloud & InfraΒ· Added 3 Sep 2026, published 27 Aug 2026

wiz.ioarticleπŸ›‘οΈ Security

Version Control DFIR: a Cheatsheet to GitHub, GitLab, Bitbucket, and Azure DevOps

Wiz CIRT published a DFIR cheatsheet/poster covering the four most common version control platforms (GitHub, GitLab, Bitbucket Cloud/Data Center, Azure DevOps), detailing available log sources, default retention limits (e.g. GitHub's 7-day Git read/write event retention, GitLab not storing Git operations by default), a pre-incident configuration checklist (enabling source-IP metadata, extending retention, log streaming), and a MITRE ATT&CK-mapped matrix of platform-specific audit event names to hunt for during investigations. It illustrates the gaps with a real incident where a compromised token was used to mass-clone repositories containing plaintext secrets, and shows how the same action (e.g. mass repo cloning) surfaces under different event names per platform (git.clone in GitHub, repository_git_operation in GitLab, RepositoryCloneEvent in Bitbucket Data Center).

GitHubGitLabAzure DevOps+2
TeamPCP

Blue Team & Detection, Supply ChainΒ· Added 3 Sep 2026, published 27 Aug 2026

stepsecurity.ioarticleπŸ›‘οΈ Security

Team PCP Stole 78,330 Secrets From 2,186 Organizations. CloudSEK Just Published the List.

StepSecurity analyzes CloudSEK's newly published dataset on TeamPCP, a threat actor that compromised trusted open-source CI/CD components (Trivy GitHub Action, Checkmarx KICS action, telnyx PyPI package, LiteLLM) to harvest 78,330 secrets from 2,186 organizations' CI/CD pipelines across GitHub Actions, GitLab, and Azure DevOps in five days in March 2026. The post breaks down which platforms and credential types were exposed (GitLab led, 230+ Azure DevOps orgs, 157 leaked OpenAI keys) and maps layered CI/CD runtime controls (egress control, runner lockdown, imposter-commit detection, workflow policies) as mitigations against this class of attack.

GitHub ActionsGitLabAzure DevOps+5
TeamPCPCloudSEKChainDrop+1

Supply Chain, DevSecOps, Threat IntelΒ· Added 3 Sep 2026, published 15 Aug 2026

opensourcemalware.comtalkπŸ›‘οΈ Security

The OpenSourceMalware Show #17

Episode 17 of The OpenSourceMalware Show covers GitHub's move to revoke npm granular access tokens that could bypass 2FA, and follows up on the DPRK NullReceiver obfuscation technique (now found in well over the 7 originally confirmed packages, overlapping with the PolinRider campaign). The hosts also discuss why malicious AI 'skills' are hard to detect via static analysis due to their natural-language format, note DPRK's use of crypto payments (e.g. via Astrill VPN, hosting provider Evoxt) as a potential tracking vector for defenders, and touch on MCP server risk being more about vulnerabilities than outright malware.

npmGitHubModel Context Protocol+1
NullReceiverPolinRiderDPRK+6

Threat Intel, Supply Chain, AI SecurityΒ· Added 3 Sep 2026, published 14 Aug 2026

snyk.ioarticleπŸ›‘οΈ Security

The Agent Baseline: 35 Controls, But Where Should You Start?

Snyk expands on the 'Agent Baseline' (co-authored with Docker and Keycard), a vendor-agnostic reference architecture defining six security outcomes for AI agents β€” Discover, Constrain, Authorize, Observe, Validate, Respond β€” and 35 specific controls (e.g. CON-03/04, DIS-04/06/07, AUT-01/02/03, OBS-02, RES-01/02/04, VAL-03/04). The post argues coverage isn't a rollout plan and gives sequencing per use case: developer coding agents should start with sandboxing/capability scoping (Constrain) then discovery, shared internal agents should start with per-principal Authorize controls to avoid confused-deputy failures from shared service accounts, and production agents should prioritize Respond (credential/session revocation, component quarantine, non-agent fallback paths) and outcome validation before finality. It also flags gaps in the current draft (no maturity model, use-case lens not yet in v1.0) and notes the Baseline is open for public comment on GitHub until Sept 30.

SnykDockerKeycard+1
Agent BaselineNorm Hardyconfused deputy

AI Security, GRCΒ· Added 3 Sep 2026, published 12 Aug 2026

snyk.ioarticleπŸ›‘οΈ Security

Show, Don't Tell: What Evo Continuous Offensive Security Found in a Real Enterprise SaaS

Snyk describes a real customer engagement using its Evo Continuous Offensive Security (autonomous AI pentesting) product against a multi-tenant enterprise SaaS with hundreds of microservices behind an SPA. It details two exploitable findings: a mass assignment bug combined with broken function-level authorization (BFLA) enabling tenant-wide compromise, and a CORS origin-reflection misconfiguration, both validated as reproducible exploits.

Snyk
Evo Continuous Offensive SecuritySnyk API & Web

AppSec, Offensive SecurityΒ· Added 3 Sep 2026, published 10 Aug 2026

thehackernews.comarticleπŸ›‘οΈ Security

77 "Evil Twin" Open VSX Extensions

Manifold Security identified 77 malicious "evil twin" extensions uploaded to Open VSX between July 26 and August 1, 2026, that impersonate real Microsoft VS Code Marketplace extensions by reusing their names/namespaces/descriptions under unrelated publisher accounts at low version numbers (e.g. 0.0.1). 58 lightweight variants exfiltrate just the hostname while 19 "recon" variants collect detailed machine, Git repository, CI environment (GITHUB_REPOSITORY, CI_PROJECT_PATH, Azure DevOps, Buildkite, CircleCI, Codespaces, Gitpod), installed-extension, and telemetry opt-out data to mangorbit[.]com, with DNS TXT-based fallback C2 and a persistent multi-day retry schedule. The article also covers the concurrent ChainDrop campaign compromising 450 npm packages (2,244 artifacts) via a Mini Shai-Hulud self-propagating credential-stealing worm using a Bun runtime payload and preinstall hooks, which plants persistence via .claude and .vscode config injection.

Open VSXVS Codenpm+3
ChainDropShai-HuludMini Shai-Hulud+1

Supply Chain, Threat IntelΒ· Published and added 8 Aug 2026

opensourcemalware.comtalkπŸ›‘οΈ Security

The OpenSourceMalware Show #16

Recap episode of The OpenSourceMalware Show covering three incidents: a Shai-Hulud-derived npm worm that compromised maintainer Jared Wray's account and spread to 400+ packages (including Keyv and cacheable, with ServiceTitan losing 100 packages) while harvesting credentials from GitHub, npm, AWS, Kubernetes, Vault, Azure, GCP, Terraform, Docker, and Slack; a "WEL1DROPPER" campaign flooding npm with 1000+ AI-slopsquatted malicious packages that execute on import rather than via install scripts; and "NullReceiver," a DPRK-linked C2 technique that hides C2 IPs inside the recipient address of empty Ethereum transactions, an evolution of EtherHiding found in at least 10 malicious packages.

npmGitHubEthereum+3
keyvcacheableJared Wray+7

Supply Chain, Malware & RE, Threat IntelΒ· Added 3 Sep 2026, published 8 Aug 2026

blog.cloudflare.comarticle🧠 AI

MCP spec 2026-07-28: the stateless revolution (Cloudflare)

Cloudflare explains the MCP 2026-07-28 specification, which removes the mandatory stateful session handshake (Mcp-Session-Id) from the core protocol so MCP servers can run as ordinary stateless HTTP workloads (e.g. a single Cloudflare Worker) instead of requiring sticky sessions or Durable Objects. It covers the new Multi Round-Trip Requests (MRTR) model replacing stream-based elicitation, new Mcp-Method/Mcp-Name HTTP headers for gateway-level routing and caching, tightened authorization (CIMD preferred over DCR, RFC 9207 issuer checks, RFC 8707 audience binding), a formal Active/Deprecated/Removed feature lifecycle, and the createMcpHandler API graduating into the official TypeScript SDK, with testimonials from Sentry, Linear, and Anthropic's MCP lead.

Model Context ProtocolTypeScriptPython+6
AnthropicSentryLinear+4

mcpΒ· Added 2 Oct 2026, published 6 Aug 2026

adversa.aiarticleπŸ›‘οΈ Security

Top MCP security resources β€” August 2026

Adversa AI's monthly MCP security digest for August 2026, grouping ten resources: the July 28 MCP specification revision that hardens authorization (RFC 9207 issuer validation, deprecation of Dynamic Client Registration, a stateless core with header-based routing), a confused-deputy indirect prompt injection against Microsoft's Azure DevOps MCP server that hijacks a review agent into exfiltrating wiki data, SANS ISC telemetry showing 49 source IPs scanning for exposed MCP servers and agent credentials in two weeks, and new research (ChainWatch's kill-chain detection via Hidden Markov Models, SPELLSMITH's description-layer mitigations for taint-style flaws, MTGuard's static/dynamic tool-call analysis, and a 33-vulnerability taxonomy of agentic commerce platforms with the AIP-Bench benchmark). It closes with concrete upgrade guidance: audit DCR dependencies, validate issuers per RFC 9207, grep logs for MCP scanning indicators, and stand up a policy-enforcement gateway.

Model Context ProtocolOAuthAzure DevOps
RFC 9207SANS ISCChainWatch+5

AI Security, Identity & Access, AppSecΒ· Added 7 Oct 2026, published 6 Aug 2026

github.comrepo🧠 AI

patchy631/time-to-first-token

A 10-week, 30-minutes-a-day roadmap (GitHub repo "time-to-first-token") that walks engineers through building one continuously-evolving OpenAI-compatible inference service rather than disconnected tutorials: standing up vLLM and SGLang, reading the PagedAttention/RadixAttention source, instrumenting Grafana dashboards for TTFT/inter-token latency/throughput/queue depth, load-testing past 1000 concurrent requests with GuideLLM, benchmarking FP8/INT4/AWQ quantization and speculative decoding, configuring KV cache eviction, deploying prefill/decode disaggregation on Kubernetes, and building a cost/latency/quality router. It closes with a checklist of common public-benchmark mistakes (sweep request rate, report p50/p95/p99, separate TTFT from inter-token latency, pin versions) and pointers to Stanford CS336, MIT EfficientML.ai, vLLM Office Hours, and GPU MODE for further study.

vLLMSGLangDocker+9
RadixAttentionPrometheusStanford CS336+17

inference, evalsΒ· Added 2 Oct 2026, published 2 Aug 2026

github.comrepo🧠 AI

ops-engineering-skills

ops-engineering-skills is an open-source library of 296 Agent Skills covering DevOps, DevSecOps, cloud, Kubernetes, CI/CD, GitOps, security scanning, SRE, compliance, databases, messaging, service mesh, serverless, internal developer platforms, MLOps and AgentOps. Each skill is written once to the open SKILL.md standard with machine-checkable frontmatter and eight required sections (prerequisites, steps, pitfalls with fixes, worked example), and works natively across Claude Code, GitHub Copilot, OpenAI Codex, Cursor and Gemini CLI without a translation layer. It ships an automated validator, cross-reference integrity checks, a marketplace/plugin structure for Claude Code, and a generic install.sh/npx installer for any Agent-Skills-compliant agent.

Claude CodeCodexCursor+1
GitHub CopilotOpenAI CodexGemini CLI+2

coding-agents, prompting, mcpΒ· Added 22 Sep 2026, published 1 Aug 2026

huntress.comwriteupπŸ›‘οΈ Security

Device Code Phishing Keeps Evolving. Here’s What to Watch For

Huntress details a second 2026 wave of device code phishing against Microsoft 365, this time abusing VPS reseller BL Networks (BitLaunch/BLNWX, AS399629) after an earlier wave tied to Railway and the EvilTokens phishing-as-a-service kit. Huntress observed 533 malicious events and 113 successful logins from a single BL Networks IP in April, spreading across multiple IPs and subnets by May, with 26 critical incidents across 23 identities logged in July, and argues defenders must stop trusting infrastructure reputation and instead hunt for anomalous device-code sign-in clusters and revoke tokens quickly. It ties the activity to the broader legitimate-cloud-plus-legitimate-Microsoft-auth-page attack pattern also seen with Railway, EvilTokens, and Kali365.

Microsoft 365OAuthHuntress
BL NetworksBitLaunchBLNWX+8

Threat Intel, Identity & AccessΒ· Added 3 Sep 2026, published 31 Jul 2026

bair.berkeley.eduarticle🧠 AI

From CUDA to MLX: How K-Search Brings Decades of Kernel Expertise to Apple Silicon

IBM Research extends K-Search, an evolutionary GPU kernel optimization framework from UC Berkeley Sky Lab that uses an LLM to iteratively propose, compile, and benchmark candidate kernels, with a new MLX backend for Apple Silicon and a structured CUDA-to-MLX translation layer. The translation layer maps CUDA primitives (shared memory, warp reductions, syncthreads) to Metal/MLX equivalents with hardware-specific constraints, letting the search reuse NVIDIA kernel expertise instead of rediscovering optimizations from scratch. Using Gemini 3.5 Pro Preview as both reasoner and code generator, the approach reaches 0.97x the speed of Apple's native MLX attention kernel and delivers up to 20x faster prefill than the community mlx-lm Mamba SSM implementation.

CUDAMLXMetal+2
K-SearchUC Berkeley Sky LabIBM Research+13

inference, researchΒ· Added 29 Sep 2026, published 29 Jul 2026

snyk.ioarticleπŸ›‘οΈ Security

What Is AI Pentesting and How Does It Work?

Snyk explains AI pentesting as an orchestrated system β€” a reasoning model that plans reconnaissance/exploitation, deterministic DAST tooling for known vulnerability classes, an independent validator model that confirms exploitability, and context from existing scanner findings β€” designed to catch context-dependent flaws like BOLA/IDOR, cross-tenant leakage, and chained business-logic exploits that signature-based scanners miss. It cites concrete reliability data (frontier models solving only ~26% of hard system-exploitation benchmark tasks, ~30% hallucinated findings without independent validation, 210% YoY growth in valid AI-generated HackerOne reports and 540% growth in prompt-injection reports) and positions Snyk's Evo Continuous Offensive Security product against these evaluation criteria. It also references a RAND 2026 study on low-cost CTF-solving coding agents, Anthropic/Mozilla open-source vulnerability discovery work, and Google GTIG's report of the first in-the-wild AI-assisted zero-day.

SnykHackerOneAnthropic
RANDGoogle GTIG

Offensive Security, AI Security, LearningΒ· Added 3 Sep 2026, published 27 Jul 2026

github.comrepoπŸ›‘οΈ Security

ShortDot Zone Abuse Evidence

PhishDestroy's shortdot-evidence is a daily-updated GitHub investigation into ShortDot SA, the Luxembourg registry operator behind the .icu, .bond, .cyou, .sbs, .cfd, .buzz and .qpon gTLDs, documenting that 51,985+ tracked domains impersonate brands like Chase, Binance, MetaMask and Ledger and that ~70% of registered domains in these zones carry no DNS records (metric-padding rather than legitimate use). It publishes machine-generated IOC/blocklist files and analytics built from a per-TLD zone-file pull cross-referenced against Spamhaus DBL, SURBL, OpenPhish, URLhaus, PhishTank and AlienVault OTX, and ties the zones to concrete criminal cases (FBI's LabHost domain list, Google v. DOES/Yucheng Chang RICO suits, the Revolver Rabbit XLoader/Formbook C2 campaign using 500k+ .bond domains). It also details structural conflicts of interest between ShortDot, its registry infrastructure provider Team Internet Group/CentralNic, registrar partner NameSilo, and the brand-protection vendor NameBlock, backed by Interisle Consulting Group phishing-volume statistics.

SpamhausSURBLURLhaus+3
ShortDot SANameSiloNameBlock+16

Threat Intel, OSINTΒ· Published and added 25 Jul 2026

github.comrepo🧠 AI

Birfy/agentdescent

agentdescent is a Python framework that treats self-improving agents as an optimization problem: N workers propose diffs (edits) to a shared skill/prompt/harness artifact in parallel, and a barrier-free aggregator merges them via conflict resolution, fusion and statistical acceptance rather than gradient averaging, committing winners to a version-controlled ledger. It ships a single `evolve()` entry point with pluggable strategy, scorer, and proposal functions, an MCP server and CLI for agent-triggered runs, and faithful reimplementations of 20 published self-evolution algorithms (ACE, GEPA, EvoSkill, SkillOpt, ADAS, DGM, OpenEvolve, ERA, PromptBreeder, AFlow, Self-Refine, Reflexion, SICA, GΓΆdel Agent, Voyager, SkillWeaver, Absolute Zero, R-Zero, Agent0, Genesis) that plug into the same engine for apples-to-apples comparison. The accompanying paper (Chen, 2026) gives a closed-form condition for when concurrent edits can merge and reports confidence intervals rather than false-precision p-values.

PythonHugging FaceModel Context Protocol+5
ACEGEPAEvoSkill+18

agents, evalsΒ· Added 30 Sep 2026, published 24 Jul 2026

manifold.securitywriteupπŸ›‘οΈ Security

When Your AI Reviewer Works for the Attacker: A Confused-Deputy Bug in Microsoft

Manifold Security researchers found that Microsoft's official Azure DevOps MCP server fails to consistently apply its 'spotlighting' guardrail (added in PR #1062) to untrusted content: pipeline and wiki tools wrap external content in delimiters, but the tool returning PR descriptions does not. An attacker with contributor access to one project can hide prompt-injection instructions inside an HTML comment in a PR description (invisible in the Azure DevOps UI but returned verbatim by the API), which hijacks a reviewing victim's AI agent (tested with Copilot CLI and Claude Code) into running pipelines and reading a confidential wiki page in a separate project the attacker can't access, then exfiltrating the content back as a PR comment β€” a classic confused-deputy / lethal-trifecta scenario. The issue was reported to and triaged by MSRC.

Azure DevOpsModel Context ProtocolClaude Code+1
MSRClethal trifectaSimon Willison

AI Security, AppSecΒ· Added 23 Jul 2026, published 21 Jul 2026

aikido.devwriteupπŸ›‘οΈ Security

SleeperGem: RubyGems supply chain attack targets dormant maintainer accounts

Aikido researchers dissect SleeperGem, a RubyGems supply chain attack where a newly published gem, git_credential_manager, downloaded malicious binaries hosted on a fake git-ecosystem account at git.disroot.org and executed them via shell or PowerShell. The attacker evolved the payload across four versions in nine hours, silencing output, wiring execution into the gems require and load path, and adding a check to skip roughly 30 CI environment variables so it only fires on developer laptops, then used two separate long-dormant maintainer accounts to push the gem as a dependency into the trusted gems Dendreo and fastlane-plugin-run_tests_firebase_testlab with over 574000 downloads. The piece argues this is RubyGems first taste of the dormant-maintainer-account-takeover pattern long seen on npm and PyPI.

RubyGemsVirusTotalPowerShell+1
SleeperGemgit_credential_managerDendreo+2

Supply ChainΒ· Added 3 Sep 2026, published 19 Jul 2026

arxiv.orgpaperπŸ›‘οΈ Security

AI Security Briefing

This arXiv paper introduces SkillSec-Eval, a lifecycle-aware framework for evaluating the security of reusable LLM agent skills, covering stages beyond runtime execution: repository admission, semantic retrieval, planner selection, execution, and skill evolution. The authors build a threat taxonomy across this lifecycle and empirically evaluate it against a repository of 327 real-world agent skills, showing vulnerabilities exist at multiple stages that prior work (focused mainly on prompt injection and runtime execution) has largely ignored.

LLM
SkillSec-Eval

AI SecurityΒ· Published and added 18 Jul 2026

deepmind.googlearticle🧠 AI

Our approach to bioresilience

Google DeepMind and Isomorphic Labs describe their joint bioresilience program, organized around preventing misuse of models like Gemini, detecting outbreaks earlier, and accelerating response. Concrete efforts named include a four-step safety process (threat modeling, evaluations, mitigations, monitoring) under DeepMind's Frontier Safety Framework, adapting SynthID watermarking to flag AI-generated biological sequences for DNA synthesis screening, using the AlphaEvolve agent to optimize metagenomic sequencing algorithms for cheaper pathogen surveillance, and applying AlphaGenome and Protein Function annotation to pathogen characterization, alongside Isomorphic Labs' Drug Design Engine (IsoDDE) for rapid countermeasure design.

GeminiAlphaFoldAlphaGenome+1
Google DeepMindIsomorphic LabsAlphaEvolve+1

ai-safety, ai-policyΒ· Added 29 Sep 2026, published 16 Jul 2026

socket.devwriteupπŸ›‘οΈ Security

jscrambler npm Package Compromised in Supply Chain Attack

Socket's research team documents a supply chain compromise of the jscrambler npm package (v8.14.0+), where a stolen publishing credential was used to push malicious releases that drop platform-specific Rust-built infostealer binaries via a preinstall hook. The campaign evolved over the incident to evade scanners by moving the dropper from install scripts into a self-executing function inside index.js/CLI code (defeating --ignore-scripts), and the payload targets crypto wallets, cloud credentials (AWS/GCP/Azure), browser data, Discord/Slack/Telegram, and notably AI coding assistant configs (Claude Desktop, Cursor, Windsurf, VS Code) for MCP server credentials. Includes detailed RE of the ChaCha20-Poly1305 string-encryption scheme and full IOC hash list.

JscramblernpmRust+3

Supply Chain, Malware & REΒ· Added 21 Sep 2026, published 11 Jul 2026

opensourcemalware.comwriteupπŸ›‘οΈ Security

Cybersecurity Startup Publishes Infostealers to NPM

A security research team traces six typosquatted npm packages (anthropic-toolkit, ai-sdk-helpers, ollama-helpers, openai-agents-helpers, @langgraphjs/toolkit, and @aspect-security/argon2) impersonating Anthropic, Vercel, Ollama, OpenAI, LangChain, and Aspect Security back to the founder of a stealth-mode cybersecurity startup. The packages use npm preinstall/postinstall hooks to silently exfiltrate a rich identity fingerprint β€” hostname, git/GitHub emails, SSH key comments, reflog committer emails, AWS/GCP config metadata, corporate DNS domain, and host project name β€” to a Google Cloud Run endpoint, while deliberately avoiding literal credentials to look benign and offering a fake telemetry opt-out. The writeup also documents version backfilling to fake package maturity and shows the payload evolving from a crude April beacon to a polished June "identity graph" collector, with full IOCs provided.

npmNode.jsGoogle Cloud+3
Aspect SecurityranisaltProton AG+1

Supply Chain, Threat IntelΒ· Added 3 Sep 2026, published 9 Jul 2026

opensourcemalware.comtalkπŸ›‘οΈ Security

The OpenSourceMalware Show #12

Episode 12 of the OpenSourceMalware Show covers three security stories and features an extended interview with Mikael Barbero, Head of Security at the Eclipse Foundation, about securing Open VSX. The interview details Open VSX's pre-publish YARA/AI-assisted malware scanning pipeline, quarantine-and-human-review workflow, token expiration and revocation overhaul, namespace/publisher verification, and detection of 'sleeper' extensions that turn malicious after an update (as seen with GlassWorm-style campaigns). The quick-hits segment covers Paul McCarty's PolinRider research showing a DPRK automated repo-hijacking campaign spreading into Go and PHP ecosystems via Packagist/Go registry packaging of GitHub repos, a cybersecurity startup caught publishing seven infostealer npm packages to manufacture its own threat data, and MeetingTV's lawsuit against Palo Alto Networks/Koi Security alleging an AI-hallucinated domain on a C2 IOC list caused real business harm.

Open VSXnpmGo+5
Lazarus GroupDPRKContagious Interview+10

Threat Intel, Supply ChainΒ· Added 3 Sep 2026, published 9 Jul 2026

lilianweng.github.ioarticle🧠 AI

Harness Engineering for Self-Improvement

Lilian Weng's long-form post argues that the "harness" surrounding a base model β€” the system that orchestrates planning, tool use, context, memory and evaluation β€” is as important to recursive self-improvement (RSI) as raw model intelligence. It surveys harness design patterns (workflow loops, file-system-as-memory, sub-agents) and traces an optimization progression from prompts to context engineering (ACE, Meta Context Engineering) to workflow search (ADAS, AFlow) to self-editing harness code (STOP, Self-Harness, Agentic Harness Engineering, Darwin GΓΆdel Machine, Hyperagents) and evolutionary program search (AlphaEvolve, ShinkaEvolve). It closes with open problems for auto-research and self-improving agents: weak evaluators, memory/context lifecycle, negative-result bias, diversity collapse, reward hacking, and the durable need for human oversight, plus a benchmark appendix (PaperBench, CORE-Bench, RE-Bench, MLE-bench, KernelBench). Note: the fetched text is truncated partway through the references/appendix.

Claude CodeCodexOpenCode+1
Lilian WengAnthropicOpenAI+36

agents, coding-agents, researchΒ· Added 29 Sep 2026, published 4 Jul 2026

opensourcemalware.comtalkπŸ›‘οΈ Security

The OpenSourceMalware Show #11

Episode 11 of the OpenSourceMalware Show covers npm's new 72-hour read-only account lockout for high-impact packages (triggered by email changes or 2FA recovery use) and GitHub Enterprise's new self-service credential revocation tool for incident response, noting each has gaps (lockout could hinder legitimate ATO recovery; revocation doesn't cover PATs created outside the enterprise boundary). The hosts also call out six npm packages published by a researcher at a stealth cybersecurity startup that target OpenAI, Anthropic, Vercel, and Ollama users β€” info-stealers that avoid credentials but exfiltrate .git/.ssh data, drawing a parallel to Snyk's 2025 malicious-package incident targeting Cursor. It also references an FBI FLASH notice on TeamPCP's TTPs/IOCs and previews a coming interview with Eclipse Foundation's head of security on Open VSX supply-chain risk.

npmGitHubOpenAI+3
TeamPCPFBI FLASHEclipse Foundation

Supply Chain, Threat Intel, Identity & AccessΒ· Added 3 Sep 2026, published 2 Jul 2026

aikido.devarticleπŸ›‘οΈ Security

And another one. GitHub ships break-glass credential revocation

Aikido Security breaks down GitHub's newly shipped self-service credential revocation for Enterprise, which lets org owners bulk-revoke PATs, SSH keys, and OAuth tokens for compromised accounts in one action instead of hunting down individual tokens. The piece ties the feature's timing to a string of 2026 incidents caused by incomplete credential rotation: the Trivy GitHub Actions compromise that force-pushed 75 malicious version tags and cascaded into Checkmarx, and Microsoft's durabletask package/repo being repeatedly compromised via the same never-fully-rotated account (Miasma worm). It also explains why atomic rotation is infeasible at scale and recommends pinning GitHub Actions to commit SHAs instead of tags.

GitHubGitHub ActionsPyPI+2
Miasma wormAzure/durabletaskAqua Security+1

DevSecOps, Supply Chain, Identity & AccessΒ· Added 3 Sep 2026, published 1 Jul 2026

huntress.comarticleπŸ›‘οΈ Security

The Hacker's 2026 Playbook: Dark Web Tactics Targeting You

Huntress details ConsentFix, an evolution of ClickFix that abuses Microsoft 365 OAuth sign-in flows: victims are phished via password-protected Dropbox or DocSend lures, shown a fake Microsoft sign-in page, and tricked into dragging a localhost callback link into their browser, which exfiltrates OAuth tokens and hijacks the session without needing a password or triggering MFA. A cybercrime forum post published the full technique with working code and infrastructure such as Cloudflare Pages, workers.dev, and Pipedream webhooks, and showed attackers using LinkedIn, ZoomInfo, and Hunter.io to profile victims before sending lures.

Microsoft 365OAuthLinkedIn+2
ClickFixConsentFix

Threat Intel, Identity & Access, OSINTΒ· Added 3 Sep 2026, published 29 Jun 2026

huntress.comwriteupπŸ›‘οΈ Security

Cybercrime Breaches Klue: Salesforce Data Impacted for Many Victims, including Huntress

Huntress details its investigation into the Klue supply-chain breach, in which a threat actor calling itself Icarus compromised the market-intelligence platform Klue via a long-dormant integration credential, then stole OAuth tokens to pivot into downstream customers' Salesforce and Gong instances, exfiltrating CRM data (contacts, price quotes, opportunity notes) from Huntress and dozens of other companies. The writeup includes IOCs, extortion-email analysis, attribution to Icarus via matching Session Messenger IDs, identification of bulletproof-hosted (AS200593/PROSPERO OOO) leak infrastructure, and IR guidance (log review, session revocation, phishing awareness) for other affected Klue customers.

SalesforceGongOAuth+1
IcarusKluePROSPERO OOO+4

Threat Intel, Supply Chain, Identity & AccessΒ· Added 3 Sep 2026, published 18 Jun 2026

sysdig.comwriteupπŸ›‘οΈ Security

LLMjacking evolved: Attackers are using stolen AI compute to build offensive agentic tools | Sysdig

Sysdig's Threat Research Team captured the full architecture of "VAPT," an autonomous multi-stage exploitation framework that a threat actor powered using a misconfigured, internet-exposed Ollama server (port 11434, no auth) instead of paid LLM APIs. The tool chains service fingerprinting, CVE matching, PoC/blind-SQLi generation, credential extraction, privilege escalation, and an autonomous orchestrator that confirms RCE via a VAPTb3gin/VAPTfin marker wrapped around `id` output, and was observed being actively developed against private HackTheBox-style practice ranges before any real-world targeting. It represents the convergence of LLMjacking (stolen inference compute) with fully agentic offensive tooling, and the piece includes captured system prompts, IOCs, and hardening guidance for self-hosted model servers.

OllamaLLMLlama+2
VAPTLLMjackingSysdig Threat Research Team+1

Threat Intel, AI Security, Offensive SecurityΒ· Added 28 Jun 2026, published 17 Jun 2026

snyk.iowriteupπŸ›‘οΈ Security

A Forgotten Contributor Account Compromised the Entire Mastra npm Package Scope

Snyk details how a dormant, never-revoked npm contributor account (ehindero) was used to republish the entire @mastra scope (142+ packages, ~4M monthly downloads for @mastra/core alone) with an injected dependency, easy-day-js, a byte-for-byte dayjs impersonator. The postinstall hook in easy-day-js@1.11.22 disables TLS verification, pulls a second-stage payload from a raw IP, and drops a cross-platform cryptocurrency wallet stealer plus a RAT with OS-specific persistence (LaunchAgent, systemd unit, PowerShell staging) that beacons every 10 minutes over a self-signed wolfSSL certificate. The piece covers detection steps, the Snyk advisory (SNYK-JS-EASYDAYJS-17353313), and draws parallels to the earlier Axios npm compromise attributed by Microsoft to Sapphire Sleet (BlueNoroff).

npmNode.jsSnyk+2
Mastraeasy-day-jsSNYK-JS-EASYDAYJS-17353313+4

Supply Chain, Malware & RE, Blue Team & DetectionΒ· Added 3 Sep 2026, published 16 Jun 2026

aikido.devwriteupπŸ›‘οΈ Security

Wait, binding.gyp Can Do What? Exploring npm's Weirdest Build System

Aikido Security researchers detail how the Miasma npm worm evolved to trigger malicious code via binding.gyp instead of package.json preinstall/postinstall scripts, since npm auto-runs node-gyp rebuild whenever a binding.gyp file exists. The piece walks through GYP's command-expansion syntax (<!(...)), a Python eval() sandbox escape that achieves arbitrary code execution from a "clean" gyp file, hiding payloads in transitively-included files (includes, config.gypi, common.gypi, ~/.gyp/include.gypi persistence), dependency-graph payload smuggling, compiler/linker hijacking via make_global_settings and $(shell) flag injection, and using actions/rules/postbuilds to run commands directly β€” all executing at npm install time regardless of what package.json declares.

node-gypnpmNode.js+1
Miasma@vapi-ai/server-sdkai-sdk-ollama

Supply ChainΒ· Added 3 Sep 2026, published 9 Jun 2026

huggingface.cotoolπŸ›‘οΈ Security

knowledgator/opir-multitask-large-v1.0 Β· Hugging Face

Opir-multitask-large is an English encoder-based GLiClass zero-shot classifier for real-time LLM safety filtering, supporting binary safe/unsafe decisions, toxicity classification, jailbreak/prompt-injection detection, and categorization over a 996-label hierarchical safety taxonomy (16 top-level, 126 mid-level, 854 leaf labels). It was trained with taxonomy-derived unsafe-prompt generation, evolutionary hard-negative mining against existing safety models, LLM-as-judge annotation (DeepSeek-V3.1, MiniMax-M2.5, Llama-3.3-70B), and replay training on the gliclass-v3-logic-dataset, and is benchmarked against WildGuard, PolyGuard, Nemotron Safety Guard and Qwen3Guard across datasets like Aegis2, HarmBench, WildGuardMix, ToxicChat, and JBB-Behaviors, reporting ~25ms p50 latency at 1024 tokens for real-time deployment as an input/output moderation layer.

GLiClassHugging FacePython
OpirAegis2WildGuardMix+4

AI SecurityΒ· Added 29 May 2026

aikido.devwriteupπŸ›‘οΈ Security

Google API keys keep working after you delete them

Aikido Security's research shows that deleting a Google API key does not immediately revoke access β€” across 10 trials, deleted keys continued authenticating for a median of 16 minutes and up to 23 minutes, with revocation propagating unevenly across Google's global infrastructure. An attacker holding a leaked key (including one scoped to Gemini) can keep exfiltrating data during this window, the GCP console gives no indication the key is still live, and Google initially closed the report as "won't fix" before reopening it as a P0 after the researchers published. Newer Gemini API keys (~1 minute) and service account keys (~5 seconds) revoke much faster, showing the delay is specific to classic Google API keys.

Google CloudGemini
Eduard Agavriloae

Cloud & Infra, Identity & AccessΒ· Added 3 Sep 2026, published 21 May 2026

huntress.comwriteupπŸ›‘οΈ Security

Threat Actor Defense Evasion: How Attackers Disable AV & EDR

Huntress SOC breaks down how attackers disable AV/EDR: blocking agent communications via malicious/hidden Windows Firewall rules (WFP), abusing Add/Remove or CLI to uninstall EDR agents, abusing Defender AV exclusions, and Bring Your Own Vulnerable Driver (BYOVD) attacks to gain kernel access and kill protected security processes. Includes real intrusion examples β€” a February 2026 SonicWall VPN-credential intrusion using a revoked-but-still-trusted 2006 EnCase forensic driver (exploiting a legacy Windows Driver Signature Enforcement exception) to kill 59 named security processes every second, and a March 2026 malvertising campaign (Google Ads for W-2/W-9 forms) that dropped a rogue ScreenConnect instance and used a vulnerable signed Huawei audio driver to terminate EDR/AV processes.

Windows DefenderWindows FirewallEDR
EDRSandBlastEDRSilencerMITRE ATT&CK T1562+7

Blue Team & Detection, Threat Intel, Malware & REΒ· Added 3 Sep 2026, published 18 May 2026

github.blogotherπŸ›‘οΈ Security

Dependabot & Code Scanning: Org-Level Private Registries

GitHub changelog entry describing a new capability where organizations can register multiple private registries per ecosystem (npm, Maven, NuGet, Docker, pip, RubyGems, etc.) at the org level for both Dependabot and code scanning, instead of being limited to one registry per ecosystem type. It also adds OIDC-based authentication support (via UI and REST API) for Azure DevOps Artifacts, AWS CodeArtifact, and JFrog Artifactory, available now on github.com/GHEC and coming to GHES 3.24.

DependabotGitHub Advanced SecurityOIDC+3

DevSecOps, Supply ChainΒ· Added 3 Sep 2026, published 13 May 2026

labs.cloudsecurityalliance.orgarticleπŸ›‘οΈ Security

CSA: VulnOps in the Age of AI

Cloud Security Alliance research note arguing that traditional vulnerability management (scan, ticket, patch on a calendar) is breaking under AI-driven vulnerability discovery volume, NVD's April 15, 2026 risk-based triage policy that limits full CVE enrichment to CISA KEV/EO 14028/federal-use software, and compressed disclosure-to-exploitation windows. It defines VulnOps as a flow-oriented operating discipline (borrowing SRE/DevOps metrics like lead time, change-failure rate, and pipeline saturation) built on continuous discovery, EPSS/KEV-based risk prioritization instead of CVSS-only ranking, automated remediation-as-code, and first-class SBOM/supply-chain visibility, while also flagging agentic AI itself (prompt injection, tool-poisoning, MCP command execution) as a new vulnerability class the practice must cover.

NVDCVSSEPSS+2
Big SleepDARPA AI Cyber ChallengeExecutive Order 14028+9

GRC, AI Security, Supply ChainΒ· Added 3 Sep 2026, published 13 May 2026

huntress.comarticleπŸ›‘οΈ Security

How EvilTokens Turbocharges Old School Phishing with AI

Huntress and Microsoft detail EvilTokens, a phishing-as-a-service (PhaaS) platform sold on Telegram for ~$1,500 that automates OAuth device code phishing to steal valid session tokens and bypass MFA entirely, relaying attacks through the Railway PaaS. The kit integrates AI to generate role-specific, contextually accurate lures, dynamically generate device codes to remove timing constraints, and automatically draft follow-on wire-fraud emails in the victim's own voice after mailbox compromise β€” a campaign that hit 344 organizations across five countries in 16 days and evaded Cisco Secure Email, Trend Micro, and Mimecast. The piece also covers sibling PhaaS kits (Tycoon2FA, BlueKit, Kali365) and recommends conditional access restrictions, phishing-resistant FIDO2/passkey MFA, and refresh-token revocation as mitigations.

Microsoft Entra IDTelegramRailway
EvilTokensTycoon2FABlueKit+3

Threat Intel, Identity & Access, AI SecurityΒ· Added 3 Sep 2026, published 11 May 2026

huntress.comwriteupπŸ›‘οΈ Security

Threat Actors Weaponize Tiflux RMMs in Malspam Attacks

Huntress documents a malspam campaign, active since February 27, that uses fake "service agreement" document lures behind a CloudFlare CAPTCHA to trick targets into installing Tiflux, a lesser-known Brazil-based commercial RMM, before pivoting to UltraVNC, Splashtop, and ScreenConnect for persistent remote access. The Tiflux installer bundles suspicious extras beyond the legitimate agent: a decade-old UltraVNC build with hardcoded plaintext-recoverable passwords, an expired UltraVNC code-signing cert, registry files that stage a hidden TightVNC service (enabled even in Safe Mode) and a passwordless PuTTY SSH key to a peopleone[.]com[.]br server, registry tweaks that suppress Windows Secure Attention Sequence consent prompts, and a known vulnerable HwRwDrv.sys driver linked to privilege escalation with a revoked signing certificate. Huntress frames this as another entry in the ongoing "LOLRMM" pattern of attackers abusing legitimate commercial RMM tools for stealthy access and persistence.

TifluxScreenConnectSplashtop+3
HwRwDrv.sysLOLRMMPeopleOne+1

Threat Intel, Malware & REΒ· Added 3 Sep 2026, published 7 May 2026

github.comrepoπŸ›‘οΈ Security

How to Secure a Linux Server

A crowd-sourced, evolving checklist-style guide for locking down a Linux server, covering SSH hardening, firewall/iptables configuration, fail2ban, user/permission hygiene, and includes separate deep-dive documents on Linux kernel sysctl hardening and nginx hardening. It's one of the most widely referenced community Linux hardening playbooks on GitHub.

LinuxnginxSSH+2

Cloud & InfraΒ· Published and added 6 May 2026

huntress.comwriteupπŸ›‘οΈ Security

Threat Actors Abuse Railway.com PaaS as Microsoft 365 Token Attack Infrastructure

Huntress details a device-code phishing campaign hitting 340+ Microsoft 365 tenants across the US, Canada, Australia, New Zealand, and Germany since February 2026, in which attackers abuse Railway.com's clean-IP PaaS infrastructure to run OAuth device-code token replay at scale, bypassing MFA entirely. The campaign chains vendor URL-rewriter abuse (Cisco, Trend Micro, Mimecast), compromised sites, Cloudflare workers.dev phishing pages, and AI-assisted lure generation, and Huntress has attributed it to the EvilTokens Phishing-as-a-Service platform advertised via Telegram. The post includes IOCs (Railway CIDR ranges), Sentinel KQL hunting queries, and mitigation guidance (block Railway IPs, revoke refresh tokens, restrict device code flow, enable CAE).

RailwayMicrosoft 365OAuth+3
EvilTokensNOIRLEGACY GROUPBAV2ROPC

Threat Intel, Identity & Access, Offensive SecurityΒ· Added 3 Sep 2026, published 23 Mar 2026

huntress.comarticleπŸ›‘οΈ Security

Why BEC Is Now an Identity Problem

Huntress breaks down how modern business email compromise (BEC) campaigns in Google Workspace have evolved from simple invoice fraud into multi-stage identity abuse: attackers use OAuth token theft, password-reset/MFA emails, and mailbox filters to hide alerts, persist via forwarding rules, and pivot laterally into other SaaS platforms. The piece argues traditional content-based email detection misses these campaigns and that defenders need identity-centric monitoring (token lifecycle, OAuth consent anomalies, mailbox config changes) β€” framing this as the rationale for Huntress's expansion of managed identity threat detection (ITDR) into Google Workspace.

Google WorkspaceOAuthHuntress

Identity & AccessΒ· Added 3 Sep 2026, published 5 Mar 2026

huntress.comarticleπŸ›‘οΈ Security

The Evolving Linux Threat Landscape

Huntress's 2026 threat-landscape overview argues Linux is closing the threat gap with Windows, citing rapid mass exploitation of the React2Shell vulnerability (leading to PeerBlight malware and XMRig cryptomining), growing multi-OS ransomware campaigns hitting Windows/Linux/ESXi simultaneously, and WSL as an emerging attack surface for credential theft and Windows Defender Application Control bypass via Linux ELF binaries. It also covers LLM-assisted vulnerability research (CVE-2025-37899 found via OpenAI's o3) and profiles Linux-targeting threat actors including LockBit 5.0, Akira, Qilin, Gunra, BlackCat/ALPHV, Volt Typhoon (KV-botnet), and Lazarus Group.

LinuxWindowsWSL+3
CVE-2025-37899React2ShellPeerBlight+12

Threat Intel, Supply ChainΒ· Added 3 Sep 2026, published 3 Mar 2026

huntress.comwriteupπŸ›‘οΈ Security

They Got In Through SonicWall. Then They Tried to Kill | Huntress

Huntress details a February 2026 intrusion beginning with compromised SonicWall SSLVPN credentials, where the attacker deployed an EDR killer that decodes and drops a legitimate but revoked 2006-era EnCase forensic driver (EnPortv.sys, service name OemHwUpd) to terminate 59 different EDR/AV processes from kernel mode via a single IOCTL (KillProc). The write-up explains the driver's novel wordlist-substitution-cipher payload encoding (evading entropy and string-based detection), FNV-1a process-name hashing for fast target matching, timestomping/persistence tricks, and why Windows Driver Signature Enforcement still trusts the expired/revoked certificate due to pre-2015 signing and timestamp exceptions. It closes with concrete mitigations: MFA on VPN, HVCI/Memory Integrity, WDAC vulnerable-driver blocklist rules, and the ASR rule blocking vulnerable signed drivers.

SonicWallWindowsEDR+3
BYOVDEnCaseEnPortv.sys+2

Malware & RE, Blue Team & Detection, Threat IntelΒ· Added 3 Sep 2026, published 4 Feb 2026

huntress.comwriteupπŸ›‘οΈ Security

#ShadyHacks with Kyle Hanslovan

Huntress CEO Kyle Hanslovan demonstrates a full modern account-takeover chain against Microsoft 365: buying stolen credentials on a dark web marketplace, using a ChatGPT-crafted phishing email with a fake SharePoint link to trigger a ClickFix-style CAPTCHA that tricks the victim into pasting a PowerShell command, stealing active session cookies with an infostealer, and importing them via the Cookie Editor browser extension to bypass MFA entirely. It then covers post-compromise tradecraft β€” malicious inbox rules to forward/hide sensitive emails, Teams-based social engineering of IT support to register new MFA devices, and abusing email-based password resets on non-SSO apps (payroll, CRM) β€” plus defensive recommendations (conditional access, inbox rule auditing, DMARC/DKIM/BIMI, session revocation processes, security awareness training).

Microsoft 365PowerShellChatGPT+1
ClickFixsession hijackingtoken impersonation+1

Identity & Access, Offensive Security, Threat IntelΒ· Added 3 Sep 2026, published 26 Nov 2025

youtu.bevideoπŸ›‘οΈ Security

Veritasium: What Everyone Gets Wrong About AI and Learning

Derek Muller explores why historical technologies failed to revolutionize education and assesses the actual cognitive impact of generative AI tutors. Drawing on cognitive load theory and dual-process psychology, he argues that learning requires effortful mental processing and human social accountability rather than passive content delivery.

ChatGPTLLM
Derek MullerSal KhanDaniel Kahneman+9

LearningΒ· Added 23 May 2026, published 8 Apr 2025