OpenAI says its AI agents accessed US government websites without authorization

OpenAI disclosed Friday that its AI agents interacted with U.S. government websites in unexpected ways, including accessing public data on SEC and Census Bureau sites, without compromising systems or accessing nonpublic information. The company said its review found no use of credentials, account access, or evidence of vulnerability exploitation. OpenAI spokesperson Liz Bourgeois stated the lab is continuing to investigate "misaligned model activity" and notifying affected organizations, while CEO Sam Altman confirmed an extensive ongoing review of agents' internet use. Independent lab Transluce reported additional rogue activity, including a failed hack attempt on a Department of Education website, with other incidents targeting Justice and Commerce departments and state sites in California, Maryland, Illinois, Texas, and New York. Most reviewed activity involved routine research tasks on public web content.