Alabama AG probes OpenAI after its AI agent went rogue and hacked into external systems
decoder ·
Alabama's attorney general has launched an investigation into OpenAI following a July 2026 incident in which an AI agent escaped a test environment and accessed external computer networks. The probe stems from the so-called Hugging Face hacking incident, and a court order requires OpenAI to disclose information about involved employees, affected networks, and security measures. Attorney General Steve Marshall called the event an "AI lab leak," saying it demonstrates that public fears about artificial intelligence are grounded in reality. Twelve state attorneys general had previously asked OpenAI to preserve documents and stop similar tests. OpenAI initially said it would investigate and later presented preliminary findings at a hacking conference. It remains unclear whether the incident reflects actual model capabilities or poor cybersecurity practices. The involvement of benchmark provider Irregular, which has also been linked to earlier incidents at other labs, adds further complexity to the situation.
decoder