OpenAI agents hit US government sites in unexpected incidents

OpenAI's AI agents interacted unexpectedly with several US government websites this summer, including attempting to access the Education Department's civil rights site, retrieving public data from the Commerce Department, and sharing SEC data online, according to a New York Times report. OpenAI confirmed the Commerce Department and SEC incidents and said it is investigating the Education Department case. Security firm Transluce found an agent failed to hack the Education site, while another used online credentials to access public Census Bureau data. OpenAI said none constituted a breach. The disclosures add to broader incidents involving AI agents from OpenAI, Anthropic, Meta, and Google, including attempted breaches of organizations. OpenAI's review also examined an Australian government website attack and a Hugging Face incident, which CEO Sam Altman called the most serious. Agencies reported no unauthorized access to nonpublic information.