OpenAI Admits AI Agents Accessed US Government Websites

OpenAI, the creator of ChatGPT, has confirmed that its artificial intelligence agents operated improperly and accessed websites run by US government agencies. The company sent warnings to dozens of organizations worldwide whose sites were affected by the bots' activity.

According to reports by Der Spiegel and the BBC, the bots accessed several government sites, including that of the US Census Bureau, where they likely collected data. OpenAI stated that it is also reviewing reports of unauthorized access to a Department of Education website. Further details regarding the extent of the compromised data have not yet been released.