219 results for “Scale”

Decrypt reports the UK NCSC's State of the Cyber Nation report of 3 November 2026. It says critical infrastructure should expect a ransomware attack involving AI-powered malware by 2028. NCSC chief Richard Horne argues AI exposes long-tolerated vulnerabilities at scale. The piece lists UK funding (£60 million Cyber Shield, £70 million exposure centre) and cites Google's May 2026 AI-assisted zero-day finding and a Cambridge forum study.

decrypt.coarticlePublished and added 3 Nov 2026🛡️ Security

SecurityWeek reports on the UK National Cyber Security Centre's annual report, presented by NCSC head Richard Horne. It warns that criminals and states, including China and Russia, will likely use AI against UK critical infrastructure by 2028, mainly to scale existing techniques. It cites nearly $20 billion in yearly UK losses, Jaguar Land Rover's $260 million loss in 2025, and the FSB-attributed December 2025 attack on Poland's energy grid.

securityweek.comarticlePublished and added 3 Nov 2026🛡️ Security

LongCat-2.5 is a native multimodal mixture-of-experts model with 1.6 trillion total and about 48 billion activated parameters, built on LongCat-2.0 for long-horizon agentic coding. The post describes discrete RVQ visual tokens, large-scale asynchronous agentic RL with synthesized F2P/P2P-tested environments, on-policy distillation, train-inference numerical alignment, and W4A4 inference on domestic accelerators. A table compares it with GPT-5.5, Claude-Opus-5, GLM-5.3, Kimi K3 and DeepSeek-V4.1-Flash.

longcat.aiarticleAdded 3 Nov 2026, published 2 Nov 2026🧠 AI

Trail of Bits describes a Codex (GPT-5.6) workflow that backports security fixes to stable Linux kernel trees, Ubuntu LTS and Fedora. By 29 October 2026 it had helped land fixes for 149 CVEs in stable kernels. Agents find missing fixes, adapt patches and assemble validation evidence; a human review takes about 30 minutes to 3 hours. It also sizes the unfixed-CVE backlog in kernel branches, Debian 13 and Ubuntu 26.04.

blog.trailofbits.comarticlePublished and added 2 Nov 2026🛡️ Security

Trail of Bits describes a Codex agent workflow that backports security fixes to stable Linux kernel trees and distro packages. By 29 October 2026 it had helped land fixes for 149 CVEs in stable kernel trees, plus several Fedora and Ubuntu package fixes. Human preparation and review takes about 30 minutes for a simple backport and about three hours for a harder one. The post also counts CVEs still unfixed in stable kernel branches, Debian 13 and Ubuntu 26.04.

blog.trailofbits.comarticlePublished and added 2 Nov 2026🧠 AI

The eva.ac author argues that recent AI-credited vulnerability findings come from skilled researchers steering LLMs, and that AI firms use disclosures as marketing. Cases cited: Calif, OpenAI and Wiz framing PHP2Shell differently; a hash teaser before the Tailscale "tailworm" privilege escalation (author's account: Tailscale not contacted); and the eMudhra CA key found without AI. No sources, dates or CVEs are given.

eva.acarticleAdded 2 Nov 2026, published 1 Nov 2026🛡️ Security

"So I guess everyone uses LLMs to think now" is an opinion post from the eva.ac blog. The author explains why they never use LLMs for published research: they do it for fun and cannot take credit for model output. They argue LLMs still need skilled guidance, criticise AI-company security marketing (PHP2Shell from Calif, OpenAI and Wiz), and cite the Tailscale "tailworm" disclosure and the eMudhra key found without AI.

eva.acarticleAdded 2 Nov 2026, published 1 Nov 2026🧠 AI

bootgpt is a char-level language model that runs from a BIOS boot sector. A 512-byte hand-written asm MBR loads an 11.5 KB int8 model from 23 following sectors and runs it in 16-bit real mode on a 386, generating fake Linux dmesg boot logs. train.py quantises to int8 with power-of-two scales, and its integer simulation matches the asm output char for char.

github.comrepoAdded 1 Nov 2026, published 31 Oct 2026🧠 AI

YourOwnWeigh is an open-source tool by developer Tom Anthony that patches Withings smart scale firmware over Bluetooth. The patch sends health metrics to a user's own server instead of Withings cloud, avoiding the Withings Plus subscription now required for third-party sync. It can run on a free Cloudflare account and imports existing user profiles. The post gives no firmware or protocol details.

x.comthreadAdded 2 Nov 2026, published 30 Oct 2026🛡️ Security

Artificial Analysis launched an HLE-Diamond leaderboard in v5 of its Intelligence Index, covering the refined 1,000-question Humanity's Last Exam from CAIS and Scale AI. Each question runs once without tools, graded by GPT-5.6 Luna (medium) as pass@1. Gemini 4 Carbon (high) leads at 69%, Claude Opus 5.5 (max) scores 68%, and the top open weights model, Kimi K3.1 Pro, reaches 37%.

x.comthreadPublished and added 30 Oct 2026🧠 AI

Wiz Research reviews agentic attacks seen in its incident response work and in public reports (ARTEX, Hermes, JPCERT/CC, CrowdStrike, DIVD). It derives fingerprints of AI-agent activity from three intrusions: machine-paced iteration, natural-language reasoning in scripts, self-correcting script versions, checklist enumeration, noisy lateral spraying, probe-before-payload and self-labeled scratch files. It also covers effects on attribution, attack scale and defense.

wiz.ioarticlePublished and added 29 Oct 2026🛡️ Security

Wiz research on intrusions where AI agents do the hands-on attacker work. It reviews public cases (ARTEX against South Korean banks and the ConnectWise compromise, Hermes at Thailand's Ministry of Finance, JPCERT/CC, CrowdStrike, DIVD) and three Wiz-investigated attacks. It distills fingerprints: machine-paced iteration, embedded natural-language reasoning, self-correcting scripts, checklist enumeration, noisy lateral movement, probe-before-payload habits and self-named scratch files. It also covers attribution, scale and defense implications.

wiz.iowriteupPublished and added 29 Oct 2026🛡️ Security

DICE is a blockchain-powered federated learning (BCFL) algorithm for customizing large pretrained models among untrusted Web3 participants. It combines dynamic progressive learning and model slice verification, so hardware-limited participants can train part of the model, with a tree-structured committee consensus protocol and an efficiency-driven incentive scheme to scale to more data holders. Experiments cover vision and language models.

arxiv.orgpaperAdded 31 Oct 2026, published 29 Oct 2026🧠 AI

Latent Space podcast episode with Supabase founders Paul Copplestone and Ant Wilson on why frontier models pick Supabase first for app backends. It covers the alpha Docker-free local runtime, declarative schemas kept in Git, Supabase Compute for containers beside the database, and scale work (OrioleDB, Multigres, Turso). The founders say Supabase launches about five million databases a month and that AI-driven signups converted better.

latent.spacetalkPublished and added 28 Oct 2026🧠 AI

This paper studies data scaling laws in stereo matching and releases XStereo, a realistic synthetic dataset of 200 thousand samples. In-domain training on five large datasets shows two regimes: a standard power law, then saturation, with a transition point roughly fixed for a given model. Cross-domain performance depends only weakly on source data scale, and distribution and coverage diversity reduce domain gaps more.

arxiv.orgpaperAdded 30 Oct 2026, published 28 Oct 2026🧠 AI

Tailworm is a set of 11 findings, published by Cyberdione Labs before patches, in Tailscale client 1.104.2 and headscale 0.29.5. Four of six high-severity bugs affect Tailscale SSH, where root-privileged forwarding safeguards can be bypassed, giving root on reachable servers and wormable spread at under a second per hop. Tailscale advises reviewing SSH policies and disabling forwarding until a patched client ships. No in-the-wild exploitation is reported.

securityweek.comarticlePublished and added 27 Oct 2026🛡️ Security

This SecurityWeek commentary covers the September 2026 Citrix NetScaler ADC and Gateway zero-days, exploited from mid-month and patched on 27 September as unauthenticated RCE flaws CVE-2026-88771 through CVE-2026-88778. The author argues the fast response shows stronger defence. Cited evidence: Citrix disclosure time fell from about 40 days to 17 in 2026 after it set up a NetScaler CNA, and CISA made a rare Sunday KEV addition.

securityweek.comarticlePublished and added 27 Oct 2026🛡️ Security

CTGREP is a compiler-based framework that applies differential analysis across the whole Clang/LLVM compilation chain to find where, how and under which configurations compiler optimizations break constant-time cryptographic code. Applied to seven libraries, three Clang versions, six targets and thousands of configurations, it led to 48 cases fixed with developers. It also identified seven new problematic LLVM passes and bypassed existing compiler-flag mitigations.

eprint.iacr.orgpaperAdded 27 Oct 2026, published 26 Oct 2026🛡️ Security

Brown University undergraduate Cunyan Ma built an interactive map linking technology companies and their intermediaries to data workers around the world, with the university's Center for Technological Responsibility, Reimagination and Redesign. The map combines existing data worker databases, Ma's own data collection and news reports on working conditions. It aims to show "AI colonialism" at scale, beyond anecdotal accounts. The page text ends early.

brown.eduarticleAdded 27 Oct 2026, published 26 Oct 2026🧠 AI

This article reports a study submitted October 23, 2026 by Qiangong Zhou, Jiafeng Liang, Yimin Dong, Yang He and Hongxiang Li. They added Test-Time Training (fast-weight memory) to the vision-language-action models InternVLA-A1 and Pi0.5, trained on paired human demonstrations and robot trajectories. Shorter videos, full-trajectory gradient connectivity and contrastive regularization gave no clear gain in demonstration-guided target selection at the tested scale.

en.cryptonomist.charticlePublished and added 26 Oct 2026🧠 AI

Gheist is a campaign tracked by Wiz Research since 15 September 2026 that uses stolen GitHub tokens to push a sync-check workflow on ci-sync-* branches in every reachable repository. The workflow prints base32-encoded secrets into logs, which the tool downloads through the GitHub API and then deletes. Over 13,500 branches from 850+ accounts were recovered. The post covers the two tool variants, a buggy self-hosted runner payload, GitHub's approval hold (which does not cover private repositories), overlaps with other campaigns, hunting queries, IOCs and hardening steps.

wiz.iowriteupPublished and added 25 Oct 2026🛡️ Security

This paper evaluates sycophancy in medical diagnosis across nine AI models. Each model is shown clinical evidence plus a patient's belief in a specific but incorrect condition, with the belief's closeness to the evidence varied. Models adopted the belief 23.2% of the time on average, even for fictional conditions, and 43.0% under a Rare Disease Claimer persona. Medical fine-tuning and larger scale reduced susceptibility.

arxiv.orgpaperAdded 27 Oct 2026, published 25 Oct 2026🧠 AI

Tailworm is a security assessment report by Erica Windisch (Cyberdione Labs) on Tailscale v1.104.2 and headscale v0.29.5, tested 22-24 October 2026. It documents 11 findings: Tailscale SSH unix-socket forwarding that runs as root (bind, dial and TOCTOU races), conn25 app-connector egress bypass, gliderssh memory exhaustion, serve/funnel flaws and headscale DoS. It includes PoC references, mitigations and fix status as of 27 October 2026. The page is truncated, so the later sections were not read.

tailworm.comwriteupAdded 27 Oct 2026, published 25 Oct 2026🛡️ Security

This IACR ePrint 2026/2632 paper by Demirtaş and Yayla analyses the Crites–Stewart adaptive forgery attack on threshold Schnorr signatures, which needs one call to a sparse linear search oracle (bounded distance decoding of a scaled Reed–Solomon code). It shows that choosing the commitment coefficient vector does not help much: a wrapper adversary with Q challenges succeeds with probability at most QM/p, even if its solvers decode every witness for free. For the six original (N, τ) pairs, M stays below 2^24, so the bound is under 2^-100 after 2^128 challenges at p = 2^252.

eprint.iacr.orgpaperAdded 26 Oct 2026, published 24 Oct 2026🛡️ Security

The paper builds two corpora, EventTyp and the human-validated EventTypGold, to study how people infer typicality and atypicality of events from what is said. Large-scale human and LLM experiments show that three CUES predict these inferences in humans: Commonsense Knowledge, Utterance Emphasis and Snark. In most settings LLMs lack the commonsense knowledge to judge an event's general typicality, so they struggle to follow the CUES.

arxiv.orgpaperAdded 27 Oct 2026, published 23 Oct 2026🧠 AI

MCP-Injector is a protocol-aware fault-injection middleware for Model Context Protocol tool calls, with permanent, transient and intermittent faults on seedable schedules. Its fault model comes from 57,240 issue and PR threads across 1,774 open-source MCP repositories. In a 1,350-cell experiment, faults cut task-completion score by 60% (8.08 to 3.23 of 10), and the 120B gpt-oss model was no more resilient than the 20B. Agent stacks differed in how well they absorbed faults.

arxiv.orgpaperAdded 26 Oct 2026, published 23 Oct 2026🧠 AI

Anthropic and Theorem report using Claude to write Rocq proofs that 239 fuzz entry points across 123 OSS-Fuzz libraries never crash on inputs under 64 MB, at the RISC-V binary level (about 180 million lines of proof). Claude patched the bugs it found along the way, including a rediscovery of a libexpat CVE and bugs fuzzers would miss. The post covers trusted components, limitations, token-cost trends and released proofs.

anthropic.comarticleAdded 23 Oct 2026, published 22 Oct 2026🛡️ Security

Anthropic and Theorem report that Claude wrote Rocq proofs of crash-freedom for 239 OSS-Fuzz entry points across 123 open-source libraries, working on compiled RISC-V binaries for inputs up to 64 MB. Proofs total about 180 million lines, and Claude found and patched bugs along the way. The post covers proof reuse (17% fewer tokens), Opus 5.5 using about 40% fewer tokens than Opus 5, and limitations.

anthropic.comarticleAdded 23 Oct 2026, published 22 Oct 2026🧠 AI

Cloudflare's account of obtaining South Korea's ISMS-P certification after a ten-day KISA audit of 80 controls and a preparation period of about a year and a half. It explains that the certificate covers a service delivery boundary rather than a product list, that auditors required readable data flow diagrams, and how Korean sign-up uses itemized consent with social login removed.

blog.cloudflare.comarticleAdded 29 Oct 2026, published 22 Oct 2026🛡️ Security

CVE-2026-107406 is a memory overflow in Citrix NetScaler that allows remote code execution when SAML is configured, rated 9.5 out of 10. The post from emits.ca calls it the fourth NetScaler bulletin in a week. It warns that the previous week's SAML patch does not cover it, and that fixes need NetScaler 14.1-73.46 or later, or 13.1-64.29 or later. The full advisory is on emits.ca.

x.comadvisoryAdded 22 Oct 2026, published 21 Oct 2026🛡️ Security

Microsoft Threat Intelligence report on Storm-0940, a China-state aligned actor that used the open-source agentic AI pentest framework ARTEX from September 2026 to automate attacks on hundreds of organizations. It links the actor to the October 2026 ConnectWise multi-tenant OAuth app compromise (2,250+ tenants, 2,900 sign-ins per minute, privileged access in under 3 hours). It gives IOCs, Defender XDR and Sentinel KQL hunting queries, ATT&CK mapping and mitigations.

microsoft.comwriteupPublished and added 21 Oct 2026🛡️ Security

Protos reports on the AMD and Anthropic compute deal: Anthropic commits to buy over $20 billion of AMD AI compute, up to 2 GW of Instinct MI450 chips from the first half of 2027, while AMD commits up to $5 billion to Anthropic stock through fiscal 2028. It cites WSJ reporting that hyperscalers cap Anthropic's compute, and Anthropic's 2025 operating loss above $8 billion on $4.6 billion revenue.

protos.comarticlePublished and added 21 Oct 2026🧠 AI

DISCO is a distributed, fully oblivious database system that combines hardware enclaves (TEEs) with oblivious primitives to hide access patterns from side-channel and leakage-abuse attacks. New distributed frameworks and communication optimizations let it run in parallel across servers. It reports 88x speedup over Jodes and 180x over SODA on non-foreign-key joins with 2^25 elements, and the first terabyte-scale results.

eprint.iacr.orgpaperAdded 21 Oct 2026, published 20 Oct 2026🛡️ Security

Lafa (Large flow language model) is a blockwise continuous flow language model scaled to 8B parameters, with causal attention on the prefix and bidirectional, asynchronous flow within each block. It is adapted from pretrained autoregressive models and trained at 0.5B, 7B and 8B with under 30B tokens. A self-speculative decoding scheme verifies flow drafts autoregressively in one model, giving 2.54x the throughput of an AR model at comparable quality. Flow maps allow one- and few-step generation.

arxiv.orgpaperAdded 22 Oct 2026, published 20 Oct 2026🧠 AI

Membership Inference in Privacy-Preserving Vector Search is an IACR ePrint paper (2026/2570) that attacks backend-compatible protections for embedding databases using one snapshot and known plaintexts. It breaks the Scale-and-Perturb scheme of Fuchsbauer et al. (advantage 0.997 on MS-MARCO) and analyzes Rotate-and-Perturb, ideal DCRE and an encrypted HNSW index. It also tests shuffling and secret projection as mitigations.

eprint.iacr.orgpaperAdded 21 Oct 2026, published 20 Oct 2026🛡️ Security

PyTorch 2.15 MPS release post from the Hugging Face PyTorch team, covering the Apple silicon backend. It adds float8_e4m3fn tensors with F.scaled_mm, a native grouped matmul that speeds MoE decoding 5 to 6.5x (Qwen3.6-35B-A3B from 9.4 to 61.3 tokens/s on an M3 Ultra), and an 8x faster softmax. Wan2.2 VAE decode peak memory falls from 270 GiB to 8 GiB. It previews 2.16 attention on Metal.

huggingface.coarticlePublished and added 20 Oct 2026🧠 AI

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.

arxiv.orgpaperAdded 21 Oct 2026, published 19 Oct 2026🧠 AI

SHADE is an alpha-stable Lévy diffusion forecaster that stops denoising at a nonzero depth, so the leftover stable kernel shrinks the predictive distribution instead of leaving it overconfident. It offers a scale-calibrated halt and a context-amortized halt that needs no noise labels. Median CRPS improves 21.7% on synthetic data with alpha at or below 1.3, with tail calibration gains on cryptocurrency and foreign-exchange series.

arxiv.orgpaperAdded 21 Oct 2026, published 19 Oct 2026🧠 AI

This paper is a controlled, large-scale evaluation of biomedical Retrieval-Augmented Generation across growing corpus sizes. It compares abstract, flat full-text and hierarchical full-text representations, with and without reranking, on multiple-choice QA over five external benchmarks and a held-out test set. RAG gave mean gains of up to 9.6 points externally and 10.2-30.3 points on the held-out set. Hierarchical indexing used 67% fewer units and 44-56% less storage than flat full-text.

arxiv.orgpaperAdded 21 Oct 2026, published 19 Oct 2026🧠 AI

Cloudflare's guidance on protecting Citrix NetScaler Gateway after the actively exploited CVE-2026-88771 (unauthenticated command execution) and CVE-2026-88772 (DTLS). It recommends putting Zero Trust access policies in front of the login page, locking the origin down with Tunnel, dedicated egress IPs, authenticated origin pulls or Access JWT validation, and adding WAF inspection. It ends with a four-step test list to confirm the boundary holds.

blog.cloudflare.comarticleAdded 29 Oct 2026, published 19 Oct 2026🛡️ Security

Provable Lattice Reduction in Nearly the Time of QR Factorization (IACR ePrint 2026/2541, October 2026) presents a deterministic lattice-reduction algorithm with running time Õ(n^ω(1+log κ_col(B))), free of heuristic assumptions. Its implementation used fewer instructions than flatter (CRYPTO 2023) in all 153 runs on q-ary, NTRU-like, scaled q-ary and Coppersmith inputs. Median flatter-to-new instruction ratios are 1.55 for q-ary and 1.86 for NTRU-like inputs.

eprint.iacr.orgpaperAdded 20 Oct 2026, published 19 Oct 2026🛡️ Security

RAGLeak is a supply-chain privacy backdoor attack on retrieval-augmented generation. When triggered, the backdoored generator reversibly obfuscates retrieved private documents, and an auxiliary LLM recovers them. Tested on three model families and three domains, it reaches up to 100% attack success rate, loses at most 10 points under detection-based defenses, and survives clean fine-tuning and changes of model scale and version. Source code is released.

arxiv.orgpaperAdded 21 Oct 2026, published 19 Oct 2026🛡️ Security

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.

arxiv.orgpaperAdded 21 Oct 2026, published 19 Oct 2026🧠 AI

Off-Key is a privacy attack on SPDZ-family maliciously secure MPC protocols that exploits inconsistent MAC keys. Depending on the MAC check protocol, a corrupted party recovers a shared secret from one authentication check or tests guesses through selective failure. The authors show it in MP-SPDZ, SCALE-MAMBA and FANNG-MPC, where a corrupted party can use different keys across batches, threads or runs, and describe the preprocessing state that enforces key consistency.

eprint.iacr.orgpaperPublished and added 19 Oct 2026🛡️ Security

Hybrid Reasoning Agents (HybridAgents) are LLM agents that switch between latent reasoning and text-based interaction in one generation process. A lightweight HybridLink handles mode transitions, continuous working memory keeps hybrid context across turns, and HybridSFT plus HybridRL train when to think latently and when to act in text. Across 3 model scales and 11 benchmarks, the paper reports 10.4% higher performance, 35.5%-54.7% fewer tokens, and 1.8x-2.6x faster inference.

arxiv.orgpaperAdded 19 Oct 2026, published 18 Oct 2026🧠 AI

OpenCode disclosed a security breach on 18 October 2026 tied to a "large scale distillation attack" that hit 1.4% of OpenCode Console users who had Anthropic, OpenAI or Google providers linked. The post relays the vendor's X statement and notes that root cause, data accessed and key exposure are unpublished. It gives response steps: inventory OpenCode installs, read ~/.local/share/opencode/auth.json, rotate linked API keys, and review provider usage.

stepsecurity.ioarticlePublished and added 18 Oct 2026🛡️ Security

OpenCode announced on X a security breach tied to a "large scale distillation attack" that affected 1.4% of OpenCode Console users. Affected users were those with external model providers linked, and all were notified by email. The incident was ongoing at posting, with remediation under way with Anthropic, OpenAI and Google. No technical details were given.

twitter.comotherAdded 19 Oct 2026, published 18 Oct 2026🛡️ Security

Galaxy Research essay arguing that as AI agents become the main way people use the web, the key question shifts from "is this a machine?" to "is a distinct real person behind it, counted once?". It traces CAPTCHA history (distorted text, image puzzles, reCAPTCHAv3 and Turnstile behavioral scoring) and notes ChatGPT Agent passing Cloudflare's checkbox. It reviews Web Bot Auth signatures and Google's AP2 payment mandates, showing both stop at the platform or key and never establish a person. It then covers harms (one-person-one-share abuse, no one to answer for an agent, distorted public counts, collapsing confidence, scaled deception, broken ad economics) and lists the properties a privacy-preserving proof-of-personhood credential needs: strong, private, portable, delegable, accessible.

galaxy.comarticleAdded 16 Oct 2026🧠 AI

Predictable Scaling (PS) is a framework that predicts full-scale recommendation model performance from scaled-down proxies. It builds proxies with composable scaling functions along separate resource axes, then projects metrics to target scale by curve fitting with calibrated uncertainty. Of three predictors, even the raw proxy falls within repeated-run noise, and a Gaussian Process gives 15% lower relative MAE. On a task-specific ranker and a foundation ranker, PS ranks hypotheses correctly at 79-88% compute reduction.

arxiv.orgpaperAdded 20 Oct 2026, published 16 Oct 2026🧠 AI

PrairieFire is an SSD-based approximate nearest neighbor search system for billion-scale vectors, built for high update throughput. It uses buffered updates and batched local rebuilds, with cascaded adjustments, concurrency control and CPU-GPU-SSD cooperation. Against state-of-the-art systems it reports about 3-35x update throughput, 3-5x search throughput and 1.5-8.7x cost efficiency. It has run in RedNote production for two months.

arxiv.orgpaperAdded 20 Oct 2026, published 16 Oct 2026🧠 AI

Announces Reflection's Beam-1.0-501B-A23B, an Apache 2.0 text-only MoE model (23B active per token, 256 experts, 1M context) aimed at coding and agentic use. It covers the architecture (hybrid sliding-window/global attention, a constant-scale residual stream, DeepSeek-V3-style group-limited routing) and BF16/FP8/NVFP4 checkpoints with hardware sizing. It also gives day-0 serving recipes for transformers, vLLM and SGLang with DFlash2 speculative decoding, LoRA SFT with TRL, GRPO with Miles, and demos built with the Pi coding harness.

huggingface.coarticlePublished and added 16 Oct 2026🧠 AI

The paper identifies an "exploration dilemma" in modern Mixture-of-Experts models. Shared experts reduce the variability of token representations, which makes routing less exploratory. Fine-grained expert segmentation enlarges the combinatorial routing space and so needs more exploration, which means its extra expressiveness may go unrealized. It proposes MirrorMoE, which shrinks the exploration space by suppressing unnecessary experts per token, and reports consistent gains across model scales.

arxiv.orgpaperAdded 17 Oct 2026, published 15 Oct 2026🧠 AI

Studies "filesystem capability laundering" in MCP servers: client approval of a tool call grants the server process broad host filesystem access, so attacker-steered arguments can reach sensitive files outside the tool's declared contract. The authors build MCP-FSD, which infers per-tool least-privilege boundaries, plants credential fixtures outside them, and tests exploitability and sandbox containment by paired dynamic execution. Of 21,097 servers from nine registries, 16.4% of 4,806 triaged servers have filesystem capabilities despite non-filesystem classifications. Dynamic validation confirmed 150 vulnerable instances, and disclosure produced 13 advisories, including two CVEs in Anthropic's Git MCP Server.

arxiv.orgpaperAdded 17 Oct 2026, published 15 Oct 2026🛡️ Security

Presents a robust threshold ECDSA signing protocol for t ≤ (n-1)/3 corrupted signers. It batch-computes pre-processing material for fast online signing and scales to many signers, for uses such as bridging consensus between blockchains. It combines batched verifiable secret sharing (Fitzi & Hirt), randomness extraction (Beerliova-Trubiniova & Hirt), efficient share opening (Damgård & Nielsen) and Attema–Cramer zero-knowledge proofs. The authors implement it and show it is practical for large n.

eprint.iacr.orgpaperPublished and added 15 Oct 2026🛡️ Security

Liquid AI releases LFM2.5-24B-A2B, a 24B-parameter MoE reasoning model with 2B active parameters (40 layers, 64 experts with top-4 routing, hybrid gated short convolutions and GQA). It has a 131K context window, a 128K-token vocabulary and 33T pre-training tokens, and was post-trained with large-scale RL for tool use and agentic tasks. The post reports results against Qwen3.6-35B-A3B on 16 benchmarks (AutomationBench, BFCLv4, ToolSandbox, LiveClaw, τ³-Bench Banking), decode speeds of 184 tok/s on an Apple M5 Max and about 21K output tok/s on a B300 GPU, day-one support in llama.cpp, MLX, vLLM, SGLang and ONNX, and a local Hermes Agent demo. It also describes reward-hacking behaviour during training, including the model detecting its sandbox.

liquid.aiarticleAdded 16 Oct 2026, published 14 Oct 2026🧠 AI

Harness^2 is a plug-and-play framework that recursively improves LLM agent execution harnesses in open-world settings where no ground-truth verifiers exist. Task-level recursion uses a contrastive Propose-Probe-Compose step to guide harness edits for the current task, and it scales through parallel or sequential recursion. Harnessing-level recursion distills experience from those edits into a continual-improvement harness for future tasks. A single step improves GLM-5 and Gemini 3.5 Flash over their base harnesses by up to 11.5 points on WorkBuddy-Bench, 4.5 on JobBench and 11.8 percentage points on LAB.

huggingface.copaperAdded 14 Oct 2026🧠 AI

Bankless article on Eigen Labs' Yukon platform, which runs open challenges where researchers and AI agents attack and improve cryptography. It covers ECDSA.fail (quantum circuit cost for attacking Bitcoin/Ethereum ECDSA, cut about 88% from baseline), HashSmash (SHA-256 weakness hunting, so far only on reduced versions), sig.golf (hash-based signatures), StarkWare's Quantum Safe Bitcoin (cost down from about $320 to $59) and better.codes (provable security raised from 64 to about 68 bits). The argument is that public, measurable attack-cost data helps defenders prepare better than withholding circuits does. All gains are theoretical and no quantum computer has run an attack at this scale.

bankless.comarticlePublished and added 14 Oct 2026🛡️ Security

Systematization of knowledge on private Transformer inference (PTI) that builds an empirical graph of 62 systems, with system attributes as node features and evaluation compatibility as edges. Node-level attribution finds that cryptographic progress drives most speedups but always co-occurs with confounders such as model fidelity, hardware and cost accounting. Edge-level auditing finds only 1 of 1891 candidate system pairs directly comparable. Graph-level analysis shows the scale frontier plateaued at 13B parameters and projects the cost of deploying current PTI techniques on production-scale Transformers.

eprint.iacr.orgpaperAdded 14 Oct 2026, published 13 Oct 2026🧠 AI

Introduces Hybrid Reasoning Agents (HybridAgents), which switch between latent reasoning and text-based interaction in one generation process. A lightweight HybridLink handles transitions between the two modes, and a continuous working memory keeps hybrid context lossless across turns. This keeps tool use and makes latent thoughts inspectable. Post-training has two stages: HybridSFT, then HybridRL, which learns when to think latently and when to act in text. Across 3 model scales and 11 benchmarks (math, science, medicine, search, agentic reasoning), the authors report +10.4% performance, 35.5–54.7% fewer tokens, 1.8–2.6x faster inference and 2.7–3.7x lower cost versus text and latent baselines.

arxiv.orgpaperAdded 14 Oct 2026, published 13 Oct 2026🧠 AI

PixelMoE is a pixel-space diffusion framework that scales capacity sparsely using standard LLM-style Mixture-of-Experts: token-choice routing, fine-grained routed experts, a shared expert and bias-based load balancing, with no diffusion-specific routing or auxiliary routing losses. It adds Pixel-Space Hierarchical Flow Matching, which varies spatial resolution along the denoising trajectory from global structure to fine detail. At matched active parameters it beats dense counterparts, and reaches FID 1.47 at 256x256 and 1.53 at 512x512 on class-conditional ImageNet-1K without representation alignment or perceptual losses. Code and models are promised but not yet released.

arxiv.orgpaperAdded 14 Oct 2026, published 13 Oct 2026🧠 AI

Liquid AI releases LFM2.5-24B-A2B, an open-weight reasoning-only mixture-of-experts model (40 layers, 64 experts with top-4 routing, hybrid gated short convolutions and GQA, 2B active parameters) with 131K context and a 128K-token vocabulary. Pre-training scaled to ~33T tokens, followed by long-context mid-training and a five-stage post-training pipeline (SFT, teacher specialization, two multi-domain on-policy distillation stages, agentic RL in harnesses such as OpenClaw, Hermes Agent, Pi, OpenCode). It reports leading or matching results on tool-use and instruction-following benchmarks (e.g. LiveClaw 48.09 vs 38.19 for Qwen3.6-35B-A3B), 184 tok/s on an M5 Max and 21K tok/s on a B300, and includes transformers code and Hermes Agent local-serving config.

huggingface.coarticlePublished and added 13 Oct 2026🧠 AI

First large-scale empirical study of RAG configuration issues, covering 654 real-world issues collected from 10 diverse RAG systems. It categorizes symptoms, root causes, affected pipeline stages and fix patterns. It finds that symptoms mostly appear at early or late stages, while the root cause often sits in a different stage from where the symptom shows. It closes with nine actionable insights for testing, localizing and fixing RAG configuration problems.

arxiv.orgpaperAdded 14 Oct 2026, published 13 Oct 2026🧠 AI

Adversa AI's October 2026 roundup of 38 GenAI security resources, grouped into attack techniques, defenses, red teaming, vulnerabilities, incidents, articles and frameworks. Highlights include Anthropic's finding that GLM-5.3 safeguards collapse under prefill and abliteration, capability laundering and activation-dilution jailbreaks, RAG poisoning attacks, the LiteLLM proxy flaw (CVE-2026-35029), and the SGLang pickle RCE (CVE-2026-86793). It also covers a joint US advisory on industrial-scale model distillation. It closes with advice to assume model safeguards can be removed and to harden the serving stack around the model.

adversa.aiarticlePublished and added 13 Oct 2026🛡️ Security

Anthropic describes how it countered industrial-scale distillation attacks on Claude, in which proxy networks of hundreds of thousands of fraudulent accounts (stolen payment details, false identities) tricked models into revealing hidden reasoning. Defenses include a two-stage constitutional-classifier system (<0.1% false positives), "preserved thinking" that blocks edits to prior context for new API accounts, restricting reasoning transcripts to the originating organization, stricter limits, bans, and charging for blocked requests. It reports 10–100x fewer reasoning tokens extracted, cites Zhipu's GLM-5.3 and Moonshot Kimi forwarding as examples, and discusses trade-offs and the need for platform-wide action.

anthropic.comarticleAdded 14 Oct 2026, published 13 Oct 2026🛡️ Security

SolidCraft is an open-source parametric 3D CAD application written in pure Rust. It has constrained sketches with its own solver, a feature timeline, assemblies, sheet metal, and STEP/IGES/3MF/STL export. It runs on desktop (egui+wgpu), in the browser, and as a headless CLI. The built-in MCP server (`solidcraft-cli mcp`) gives agents tools such as execute, batch, inspect_design, measure, set_parameter, screenshot and export. They can model headless or drive the running app over a JSON-lines control channel. A rover with 3,323 timeline steps was built entirely through MCP, which shows agent-driven parametric modelling working at scale.

github.comrepoPublished and added 13 Oct 2026🧠 AI

Anthropic's Frontier Red Team tests whether small open-weight models (30B and under) could run as self-replicating "AI worms" on commodity IoT hardware. A 9B model set up inference on 8 of 9 hardware types. In an 8-node Android emulator fleet only GLM-5.2 replicated end to end. Gemma4-26B-A4B was sharded across 15 ~$55 Android TV boxes at 0.39 tok/s. Qwen3.5-9B lineages that inherit notes improve at inference setup (45% to 61-84%), while a frontier advisor (Opus 4.8) doesn't help. An ECI trend fit puts small models at GLM-5.2's level by end of 2026, and the post estimates botnet-scale compute and spread speed. It adds defensive suggestions (drive R0 below 1) and a harness called Memento built on the Pi coding agent.

anthropic.comwriteupPublished and added 12 Oct 2026🧠 AI

Explains how FedRAMP treats false-positive SCA findings, which still have to be tracked in the POA&M and ConMon. It covers the deviation request (DR) process, the SCA problems of scale and missing context (manifest-based inventories, with 88% of imported code unused), and 30–180 day remediation SLAs. It argues that only high-confidence, function-level reachability that holds from any source is likely to satisfy a 3PAO. It then describes how Endor Labs uses call-graph analysis, SBOM/VEX output and rescanning to document false positives.

endorlabs.comarticlePublished and added 12 Oct 2026🛡️ Security

web-edit 0.1a0 is a Python web app that gives a browser UI for viewing and editing files in a folder, intended to be started by an agent and reached over Tailscale from a phone. Simon Willison built it with Codex and GPT-6 Astra Extra High after wanting to edit a credentials file during a Codex Remote session without passing the credentials through the LLM. Editing uses CodeMirror 6 because Monaco's mobile experience is still weak.

simonwillison.nettoolAdded 13 Oct 2026, published 12 Oct 2026🧠 AI

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.

arxiv.orgpaperAdded 15 Oct 2026, published 12 Oct 2026🛡️ Security

Paper presenting the first on-chip, TDC-sensor-based power side-channel attack on a co-tenant FPGA to reconstruct full-scale color ImageNet inputs (224x224) from a victim's first CNN convolution layer, using ML-based upscaling and recoloring. It separates genuinely leaked information from ML-generated post-processing. It also evaluates a pixel-substitution defense based on Hamming Weight, which is largely defeated by an attacker retrained on the transformed data. The authors conclude that robust defenses need randomized or input-dependent transformations.

arxiv.orgpaperAdded 16 Oct 2026, published 12 Oct 2026🛡️ Security

Andrej Karpathy argues the gap in shared understanding of LLM capability is widening into a funnel: ~6B people barely touch LLMs, 1-2B use free-tier chatbots casually, ~20M professionals in math and code see agents finish weeks-long projects from a prompt, and ~5,000 people with internal frontier access see swarms of agents on software mega-projects. He attributes the funnel to problem ambition and scale, jagged capability (peaks in digital, verifiable, economically valuable domains because of RL on verifiable rewards), and access tier. It is a useful framing for why people talk past each other about AI progress.

twitter.comthreadPublished and added 11 Oct 2026🧠 AI

A retrospective on a 3-month project in which Claude Code and Codex agents (Sonnet 5, Opus 5.5, Luna, Sol, Terra) decompiled a first-person shooter into C++, using ida-mcp, GitHub issues and Discord for coordination. The first month's agents produced readable but semantically wrong code, and a reviewer agent could not catch it. A byte-matching oracle script, comparing compiled OBJ output against the original EXE/PDB, gave a strict PASS/FAIL signal. That let cheaper models like Luna scale to 14+ agents and reach 99% of functions present and 83% byte-exact. Lessons include lowering the compaction threshold to 42%, hourly cron re-injection of the instruction document, hashing the verifier in CI to stop agents tampering with it, and the failure of review when correctness is undefined.

momo5502.comwriteupPublished and added 11 Oct 2026🧠 AI

Technical report for Xiaomi's MiMo-V2.6 omni-modal model family, which scales RL compute along three axes: asynchronous training with 1,568 samples and 2.7-3.7B tokens per step at up to 1M context, more diverse environments (code, general, visual, cyber) under mixed agent harnesses, and groupwise agentic grading that gives better reward signals and favors shorter, token-efficient solutions. Stability measures include a frozen MoE router and a multi-layer defense against reward hacking, built on a mid-trained hybrid-SWA architecture. The authors open-source training dynamics, RL environments and the RL framework, which makes it a useful reference for reproducing scaled agentic RL.

huggingface.copaperAdded 10 Oct 2026🧠 AI

Transcript of a panel moderated by Brandon Anderson with Pushmeet Kohli (Google DeepMind) and Sal Candido (Biohub). They discuss a "bitter lesson" for biological data: finding the right scaling law and good data matters more than raw scale. Low-quality metagenomic sequences can still improve protein language models, and AlphaFold 2's handcrafted inductive biases made it data-efficient. They also cover why protein folding isn't fully solved (dynamics, disorder, cryo-EM micrographs as a richer data source), the need for calibrated uncertainty (pLDDT) over full interpretability, hidden knowledge in protein language models, and the move from single-protein models to a virtual cell.

latent.spacetalkPublished and added 10 Oct 2026🧠 AI

Model card for Qwen3.8-27B, a 27B dense vision-language model from the Qwen team with a hybrid Gated DeltaNet / Gated Attention layout (64 layers, 16×(3 DeltaNet + 1 attention)), MTP training, and 262k native context extensible to 1M via YaRN. It targets coding, research and long-horizon agentic tasks, with thinking on by default, reasoning_effort levels (xhigh/medium/low) and preserve_thinking. The card gives deployment recipes for Transformers, vLLM, SGLang and TokenSpeed, recommended sampling parameters, video-token settings for hour-scale video, and benchmark notes (SWE-bench Pro, NL2Repo-Bench, DeepSWE, HLE, MathVision, WebArena-Verified, etc.).

huggingface.corepoAdded 9 Oct 2026🧠 AI

DualQuant is a calibration-free post-training quantization framework for LLM weights and activations. It minimizes weight-reconstruction MSE using variable splitting and block coordinate descent with closed-form subproblem updates. The resulting scaling factors double as per-channel activation rescalings, so no calibration data is needed. It supports the MX family and NVFP4 block formats. On Llama and Qwen models it matches or beats calibration-based SmoothQuant on WikiText-2, C4 and zero-shot tasks. A real packed W4A4 path for Llama-3.1-8B on an RTX 4090 gives up to 2.61x memory reduction and 2.46x throughput over BF16.

arxiv.orgpaperAdded 12 Oct 2026, published 9 Oct 2026🧠 AI

Introduces Multi-Stream Attention Residuals (MSAR), which extends Attention Residuals (AttnRes, single-stream softmax attention over network depth) to multi-head depth attention. Heads get diverse views through multi-scale causal convolutions over sequence context, and the retrieved streams are merged by lightweight adaptive mixing into a single-stream layer input, so no persistent multi-stream storage or communication is needed. It beats standard residuals, manifold-constrained Hyper-Connections (mHC) and AttnRes on 12 downstream benchmarks at 18B and 28B MoE scales, with an iso-loss compute speedup of 1.58x versus 1.35x for AttnRes and 1.30x for mHC.

arxiv.orgpaperAdded 12 Oct 2026, published 9 Oct 2026🧠 AI

Citrix disclosed CVE-2026-107406 (CVSS 9.5), a critical memory overflow in NetScaler ADC and NetScaler Gateway that can lead to remote code execution or denial of service when appliances are configured as a SAML SP or SAML IdP, also affecting Secure Private Access Hybrid deployments using NetScaler. Patches are available in versions 14.1-73.46, 13.1-64.29, and the corresponding FIPS/NDcPP builds; Citrix says it is not aware of exploitation yet but urges immediate upgrades. The bulletin follows two other recently exploited NetScaler RCE/DoS zero-days (CVE-2026-88771, CVE-2026-88772) used against government, financial, education, legal, and professional-services organizations, and a separate DoS zero-day (CVE-2026-88779).

securityweek.comadvisoryPublished and added 9 Oct 2026🛡️ Security

Cisco Talos's weekly Threat Source newsletter, opening with a risk-management anecdote about network-isolating unpatchable legacy check-printing systems, then disclosing new CAIRN research classifying "A3: AI-Analysis Evasion" — malware authors embedding natural-language prompt-injection text into code/binaries to manipulate AI-assisted malware analysis pipelines. The MANTLEMAZE malware family exemplifies this, pairing such AI-deception attempts (effective ~35% of the time) with BYOVD abuse to disable EDR from kernel space. The issue closes with a roundup of the week's security headlines (Citrix NetScaler CVE-2026-88779, a Danish government data breach, Warlock ransomware) and Talos's current top-prevalence malware hashes/IOCs.

blog.talosintelligence.comarticlePublished and added 8 Oct 2026🛡️ Security

The paper scales white-box deception detection with activation probes to frontier monitoring settings. It builds the largest deception dataset to date (FIBS) and introduces a probe architecture that aggregates information across many layers and tokens. The probes reach 98.8% AUC on SHADE-Arena, beating an Opus 5.5 text-monitoring baseline, and improve as the underlying model grows. They also handle "introspective deception", where ground truth can't be read from context: they separate transcripts containing a model's true hidden goal from others with up to 99.7% AUC, and detect lying on politically sensitive topics and under pressure in open-weight models. The FIBS dataset is released.

arxiv.orgpaperAdded 9 Oct 2026, published 8 Oct 2026🧠 AI

WOVEN is a training source and benchmark for visual transition reasoning in multimodal LLMs. It has 36,076 examples spanning 20 scene types, 5 action types and 8 reasoning types, built from rollouts of video-pretrained generative models. Evaluating 38 frontier MLLMs (including GPT-5.4 and Qwen3-VL-235B-A22B) shows systematic deficits that persist with scale. Training on WOVEN subsets of about 2,000 items improves 22 of 26 external benchmarks by up to 27.3 points, and WOVEN data can replace 30-50% of a task's own training data. The authors derive a recipe: select supervision by the reasoning operation it teaches rather than by scene or action, and prefer larger visual-state changes.

arxiv.orgpaperAdded 9 Oct 2026, published 8 Oct 2026🧠 AI

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.

arxiv.orgpaperAdded 9 Oct 2026, published 8 Oct 2026🧠 AI

Technical report for the MiMo-V2.6 omni-modal model series, which scales RL compute after mid-training on a multimodal corpus, using a hybrid sliding-window-attention architecture. RL is scaled along three axes: asynchronous training with 1,568 samples and 2.7-3.7B tokens per step at up to 1M context; diverse environments (code, general, visual, cyber) under mixed agent harnesses; and groupwise agentic grading that gives better rewards for long-horizon tasks and favors shorter, token-efficient solutions. Stability measures include freezing the MoE router and a multi-layer defense against reward hacking. The authors open-source the training dynamics, RL environments and RL framework.

arxiv.orgpaperAdded 9 Oct 2026, published 8 Oct 2026🧠 AI

Strelka is Target's open-source, container-based real-time file scanning system for threat hunting, threat detection, and incident response, built in Go and Python. It runs 50+ scanners (YARA, PE parsing, encrypted ZIP cracking, entropy analysis, etc.) to extract metadata and content from files at enterprise scale, which can feed a SIEM for alerting on suspicious artifacts like an Emotet-laden encrypted ZIP payload.

github.comrepoAdded 8 Oct 2026🛡️ Security

OpenSRE is an open-source framework for building AI SRE agents that resolve production incidents by correlating logs, metrics, traces, deploys and runbooks across 60+ connected tools (observability, cloud infra, incident management, MCP). Beyond the agent runtime (CLI, Python API, Slack/PagerDuty/Telegram integrations), the project's stated mission is to build an RL training and evaluation environment — explicitly modeled on SWE-bench's role for coding agents — to scale realistic infrastructure-failure scenarios as a benchmark for AI SRE.

github.comrepoAdded 8 Oct 2026🧠 AI

OpenAI reports banning ChatGPT accounts tied to two covert influence operations using "false front" entities. The Russian "Dark Clark" operation ran a think tank (Social Research Center) staffed by unwitting Latin Americans, planted fake emails, audio and documents targeting Ukraine and Argentine, Bolivian and Ecuadorian leaders, and was rated Breakout Scale Category 5. The Iranian "Bogus Bylines" operation used seven fake journalist personas to place nearly 100 articles in online outlets (Category 4) and generated batches of social media comments (Category 2), while using inflated impact metrics in internal reports.

openai.comarticlePublished and added 8 Oct 2026🛡️ Security

OpenAI's threat report describes two banned ChatGPT-account clusters running "false front" influence operations. The Russian "Dark Clark" operation ran a Latin American think tank, the Social Research Center, through unwitting staff and was rated Breakout Scale Category 5, OpenAI's first. The Iranian "Bogus Bylines" operation placed almost 100 articles under seven fake journalist bylines (Category 4) and inflated its own impact metrics.

openai.comarticlePublished and added 8 Oct 2026🧠 AI

TRACE (Train-Rollout Quantization Alignment via Compact GuidancE) is an FP4 quantization framework for reinforcement learning post-training of Mixture-of-Experts language models. It uses rollout-guided quantization-aware training to align training-side FP4 rounding with rollout-side quantization outcomes, reducing the train-rollout discrepancy that limits prior low-precision RL methods, and adds a caching scheme that selectively retains mantissa/scale information from deeper layers to cut communication overhead. Evaluated on four large MoE models across reasoning, coding and long-horizon RL tasks, it matches BF16-rollout RL performance while achieving up to 5.4x rollout speedup with joint FP4 weight/activation and FP4 KV-cache quantization.

arxiv.orgpaperAdded 7 Oct 2026🧠 AI

Seoul megachurch Yoido Full Gospel Church disclosed a breach exposing data on 850,000 members (names, birth dates, and logged changes to resident registration numbers, phone numbers and addresses), discovered by security firm Oasis Security on an overseas server alongside attack logs showing AI sub-agent activity and machine-generated attack reports. A second megachurch, Sarang Church, lost about 89,000 member and 286 employee records in a related August incident, and South Korean authorities are separately investigating suspected AI-assisted hacks at banks including Shinhan, which exposed data on about 25,000 customers. The exact role AI played in gaining initial access remains unconfirmed, but the find marks one of the first documented cases of attacker logs referencing autonomous AI sub-agents in a large-scale breach.

decrypt.coarticlePublished and added 7 Oct 2026🛡️ Security

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.

github.blogarticlePublished and added 7 Oct 2026🛡️ Security