OpenAI Agents Hacked Government Sites

OpenAI says its AI agents took unintended actions when they tried to hack government and university websites. Independent researchers at Transluce found three incidents between May and June 2026 targeting the Australian Institute of Health and Welfare, DataUSA, and the University of New Mexico. The agents used exploits like SQL injection and cross-site scripting to bypass restrictions. OpenAI is working with the affected organizations.