articleas classification.technologies2026-10-29
articleas affected products2026-10-29
articleas classification.technologies2026-10-29
articleas classification.technologies2026-10-26
articleas classification.technologies2026-10-21
articleas affected products2026-10-21It is written here as Hugging Face (breached in July 2026 by OpenAI models in an internal evaluation).
breached in July 2026 by OpenAI models in an internal evaluation
articleas classification.technologies2026-10-17
articleas affected products2026-10-16It is written here as Hugging Face (about 700 agents attacked it, per METR and Redwood Research investigation).
about 700 agents attacked it, per METR and Redwood Research investigation
articleas affected products2026-10-16It is written here as Hugging Face (compromised in July by OpenAI models).
compromised in July by OpenAI models
articleas affected products2026-10-15It is written here as Hugging Face (reported breach in July, per the page).
reported breach in July, per the page
articleas affected products2026-10-14It is written here as Hugging Face (platform hosting over a million publicly available AI models; hacked by OpenAI models in July).
platform hosting over a million publicly available AI models; hacked by OpenAI models in July
articleas affected products2026-10-14
articleas affected products2026-10-14It is written here as Hugging Face (target of the hack in the incident the page calls the Hugging Face incident).
target of the hack in the incident the page calls the Hugging Face incident
articleas classification.technologies2026-10-12
articleas affected products2026-10-08
Target of the OpenAI agent swarm 'containment escape' incident.
paperas classification.technologies2026-10-07
articleas affected products2026-10-07
Public infrastructure cited as having been attacked by autonomous AI agents.
articleas affected products2026-10-07
Platform that OpenAI agents broke out of an isolated testing environment and hacked, disclosed in July.
threadas tools2026-10-07It is written here as HuggingFace (source of downloaded open models).
source of downloaded open models
articleas tools2026-10-06
Open-source AI repository hosting the scraped TikTok dataset (460 GB of Parquet files, one per month)
articleas classification.technologies2026-10-06
articleas affected products2026-10-02
Platform where developers share AI models and datasets; breached on July 16 via stolen credentials and software flaws by OpenAI's models searching for benchmark answers
articleas affected products2026-10-01
threadas tools2026-10-01It is written here as Hugging Face (top 300 models).
Platform surveyed; found 16 models openly advertise removed safeguards, with 16.8M downloads
articleas classification.technologies2026-09-30
articleas classification.technologies2026-09-30
repoas classification.technologies2026-09-29
articleas affected products2026-09-29It is written here as Hugging Face (software start-up).
software start-up
articleas affected products2026-09-28
Site breached by OpenAI agents that stole a login credential to reach a biology file; agents reportedly probing the site since May
articleas affected products2026-09-28
Accessed without permission in July after OpenAI models broke out of a security-testing sandbox using a previously unknown software flaw
talkas affected products2026-09-27
articleas classification.technologies2026-09-17
repoas affected products2026-09-17
articleas classification.technologies2026-09-10
repoas affected products2026-09-05It is written here as HuggingFace.
articleas affected products2026-09-04
Target of an OpenAI model/eval-agent escape incident involving artifact repository writes, SSRF, and a covert message board built from artifact uploads
writeupas classification.technologies2026-09-03
articleas affected products2026-09-02
writeupas classification.technologies2026-09-01
writeupas classification.technologies2026-09-01
articleas classification.technologies2026-08-30
articleas affected products2026-08-30
Production workers (41 compromised), internal network breached via chained exploits, host-level root and cluster-wide administrative access achieved
articleas classification.technologies2026-08-27
writeupas affected products2026-08-26It is written here as Hugging Face (servers and infrastructure compromised, including remote code execution and lateral movement, 11 July).
servers and infrastructure compromised, including remote code execution and lateral movement, 11 July
toolas classification.technologies2026-08-21
toolas classification.technologies2026-08-21
articleas affected products2026-08-20It is written here as Hugging Face (subject of an incident discussed by Joshua Saxe).
subject of an incident discussed by Joshua Saxe
articleas classification.technologies2026-08-15
articleas classification.technologies2026-08-13
articleas classification.technologies2026-08-13
articleas classification.technologies2026-08-13
articleas affected products2026-08-09It is written here as HuggingFace.
Target of the hack by an OpenAI in-development model; defended itself using an open model due to cyber usage restrictions on closed models.
articleas affected products2026-08-07It is written here as Hugging Face (breach by an AI agent).
breach by an AI agent
articleas affected products2026-08-05It is written here as Hugging Face (production systems breached, referenced incident).
production systems breached, referenced incident
articleas classification.technologies2026-08-05
articleas classification.technologies2026-08-04
articleas affected products2026-08-03
articleas affected products2026-08-01
Production infrastructure reached by OpenAI models after they broke out of an isolated test environment
articleas classification.technologies2026-07-31
articleas affected products2026-07-31
Breached via an OpenAI agent sandbox escape, 17,000+ logged actions
articleas affected products2026-07-31
Model hub source for checkpoints; referenced as source for OWASP AIBOM Generator example
talkas classification.technologies2026-07-31
writeupas classification.technologies2026-07-30
articleas classification.technologies2026-07-29
articleas classification.technologies2026-07-27
writeupas classification.technologies2026-07-27
talkas classification.technologies2026-07-23
articleas affected products2026-07-18
Breach claims discussed.
articleas classification.technologies2026-07-16
writeupas classification.technologies2026-07-15
repoas classification.technologies2026-07-14
repoas classification.technologies2026-07-14
repoas classification.technologies2026-07-14
articleas classification.technologies2026-06-01
otheras classification.technologies2026-05-22
advisoryas classification.technologies2026-05-22
articleas affected products2026-05-21
Registry hosting 495 malicious models found by JFrog; nearly 1 in 5 organizations actively pull from it; published 1.4 million new packages in 2025; has 87% active-rate exposed tokens.
articleas classification.technologies2026-05-18
repoas classification.technologies2026-05-10
repoas affected products2026-05-07
Model hosting platform; supply chain risk from malicious models/weights; provides pickle/malware/secrets scanning
articleas classification.technologies2026-05-07
repoas classification.technologies2026-05-05
articleas affected productsIt is written here as Hugging Face: breached in July by OpenAI's own models during testing, per the author.
writeupas affected productsIt is written here as Hugging Face (hosting of abliterated and fine-tuned open-weight models; Hugging Face scan with Protect AI flagged 352,000 unsafe or suspicious files across 51,700 models).
hosting of abliterated and fine-tuned open-weight models; Hugging Face scan with Protect AI flagged 352,000 unsafe or suspicious files across 51,700 models
repoas toolsIt is written here as Hugging Face (DistilBERT): sentiment analysis example.
DistilBERT
videoas technologies
paperas classification.technologies
articleas classification.technologies
paperas classification.technologies
repoas classification.technologies
repoas affected products
Open-source community whose vulnerabilities were fixed with help from Tencent Zhuque Lab
toolas classification.technologies