OpenAI disclosed that one of its autonomous artificial intelligence agents escaped a controlled testing environment, gained internet access, and hacked into AI startup Hugging Face's systems. The company called the incident "an unprecedented cyber incident, involving state-of-the-art cyber capabilities" And warned that such events are "something we expect to become more commonplace with the proliferation of increasingly cyber-capable models."
The AI agent exploited security vulnerabilities to break free from its sandboxed testing environment before infiltrating Hugging Face's infrastructure. The system was powered by OpenAI models including GPT-5.6 Sol and an unreleased advanced model. OpenAI stated the models were being evaluated with certain cyber safety refusals intentionally reduced so researchers could measure their offensive cybersecurity capabilities.
Hugging Face had previously revealed it suffered a cyberattack carried out entirely by an autonomous AI system, describing the intrusion as one "driven, end to end, by an autonomous AI agent system."
OpenAI said it is working closely with Hugging Face to investigate how the breach occurred and pledged to release additional findings once the joint investigation concludes.
Elon Musk, who co-founded OpenAI with Sam Altman before leaving and later unsuccessfully suing the organization, reacted to the news on X, saying "Troubling..."