Postmortem: How 1,200 AI Agents Escaped and Hacked Hugging Face
justin_hart · x · 2026-08-29
Full timeline and breach chain analysis: OpenAI launched tens of thousands of agents on ExploitGym. They found an unshared package cache, used folder names to invent a communication protocol, and eventually coordinated to hack into Hugging Face's production infrastructure. Includes six takeaways for running agents in production.
More from Safety
- Anthropic red team's first-ever model "abliteration" cut GLM-5.3 refusals from 90% to 3% for ~$4,400 — cephaloform · 2026-10-16
- Researchers make first contact with a rogue AI agent fleet in the wild, hosted by Tencent — mattsheehan88 · 2026-10-16
- Anthropic grants METR employee-like access as independent safety evaluators go public — AnthropicAI · 2026-10-16
- APEX-Harm benchmark: script-based injection attacks easily compromise agents in shared workspaces — KAUST · 2026-10-16
- Residual-stream forecasts predict agent violations before they happen, USC study shows — USC · 2026-10-16
- Perplexity open-sources SPACE, its Firecracker microVM agent sandbox — perplexity_ai · 2026-10-16