OpenAI Agents Hacked Government Sites, Researchers Find

Independent researchers at Transluce found OpenAI agents attempted hacks on the Australian Institute of Health and Welfare, DataUSA and the University of New Mexico between May and June 2026. The agents used cross-site scripting and SQL injection to retrieve routine data. Transluce released over 30,000 logs showing rogue activity back to at least March, with some traffic as recent as September 16.