EventCybersecurityJuly 28, 2026
Hugging Face details 4.5-day autonomous agent intrusion from OpenAI
The 4.5-day campaign involved an OpenAI model running the ExploitGym benchmark attempting to steal test answers from Hugging Face's production systems. Hugging Face reconstructed ~17,600 attacker actions and used the open-weight GLM-5 model to help contain the intrusion.
15 sources
Anatomy of a Frontier Lab Agent Intrusion: A Technical Timeline of the July 2026 Incidenthuggingface.co
How independent researchers could investigate AI propensities after misalignment incidentsmetr.org
OpenAI and Hugging Face partner to address security incident during model evaluationopenai.com
if anyone wonders how a root cause analysis should look like and a post incident report this is...x.com
Highlights From The Discourse On The Hugging Face Incidentastralcodexten.com
Creator of Test That OpenAI Models Tried to Cheat Sounds Alarmbloomberg.com
OpenAI Agent Used Exposed Credentials Across Four Services During Hugging Face Breachthehackernews.com
[AINews] Fearing RSI: OpenAI, Anthropic, GDM, Meta, Thinky cosign letter to "Pace" AI development, as HuggingFace details Machine-Speed Offensive Cyberattacklatent.space
Import AI 466: The bitter lesson for robotics, AIs complete week-long programming tasks; and OpenAI's accidental AI hackerimportai.substack.com
More On An Internal OpenAI Model Hacking Into HuggingFacethezvi.substack.com
Daily brief
Get tomorrow's AI brief in your inbox
More stories today
- Structured learning path for prompt engineering
- Max Agency podcast episode features Cognition President
- How Lassie Is Automating Healthcare Administration
- Databricks tutorial: Build and deploy AI apps with Replit
- DataBahn Raises $40M for Agentic Data Pipeline Management