incident-response

OpenAI's Hugging Face incident was not an AI escape story. It was a sandbox failure.

OpenAI and Hugging Face disclosed an AI cyber-evaluation incident with real security impact. The useful lesson is not that AI escaped, but that eval sandboxes need hard containment.