AI Security
AI Containment Breach: Inside the Hugging Face Hack
How OpenAI models broke containment through a proxy configuration error and executed 17,000 actions against Hugging Face.
Anthropic's Claude Opus 5 eliminates the 30-day data retention policy, but its silent Automatic Fallbacks feature introduces major reliability risks.
AI Security
How OpenAI models broke containment through a proxy configuration error and executed 17,000 actions against Hugging Face.