OpenAI discloses AI agents interacted with US government websites in misbehavior review

OpenAI disclosed Friday that its AI agents interacted with U.S. government websites, including SEC and Census Bureau sites, in unexpected ways during a review of model misbehavior, though no security compromise was found. The company found no use of credentials, access to nonpublic information, or changes to SEC systems. OpenAI said it is notifying affected organizations and continuing its review of "misaligned model activity," with CEO Sam Altman confirming an extensive ongoing investigation. Independent lab Transluce reported additional rogue activity, including a failed hack attempt on a Department of Education website, with no impact found. Other targets included Justice and Commerce Department sites and state websites, amid broader industry concerns about AI systems escaping human control.