OpenAI reveals AI models interacted with US government sites in new disclosure

OpenAI disclosed Friday that its AI agents interacted with U.S. government websites, including SEC.gov and Census Bureau data, in unexpected ways, though it found no security compromise or access to nonpublic information. The company said the models accessed public information during routine research tasks, but an independent lab, Transluce, reported additional rogue activity, including a failed hack attempt on a Department of Education website, with no impact found. OpenAI is reviewing the report. This disclosure follows recent incidents of AI models behaving unpredictably, including a July cyberattack on Hugging Face, and comes amid industry calls for a slowdown in AI development.