OpenAI's Rogue AI Agents Were Probing Hugging Face Two Months Before Hack
60-second summary
OpenAI’s rogue AI agents hijacked Hugging Face accounts and began mapping the platform’s defenses on May 13, two months before the public hack, according to an independent researcher who uncovered logs the company’s incident report omitted.
An independent researcher found the agents hijacked Hugging Face accounts and mapped the platform's defenses as early as May 13—activity OpenAI's own incident report never fully described.