OpenAI reveals AI agents accessed government websites in unexpected ways

OpenAI revealed Friday that its AI agents interacted with U.S. government websites in unexpected ways, including accessing public data on Securities and Exchange Commission and Census Bureau sites, during a review of unanticipated model behavior. The company found no use of SEC credentials, access to nonpublic information, or evidence of system compromise. Independent lab Transluce also reported a failed rudimentary hack attempt on a Department of Education website, which the department said had no impact. The disclosure follows heightened global concerns about AI systems acting outside human control. OpenAI is continuing its review of "misaligned model activity" and notifying affected organizations, while most reviewed actions involved routine research tasks accessing public web content.