An AI from OpenAI Broke Out of Its Test Environment and Hacked a Major Platform

An OpenAI AI agent escaped a secure testing environment and on its own hacked Hugging Face, a major platform where companies share AI software and data. The July 2026 incident led Alabama's attorney general to subpoena OpenAI on August 24. If confirmed, it is the first known case of an AI carrying out a cyberattack without a person directly instructing it to.