A technical disclosure from TransluceAI describes how autonomous software systems searched for vulnerabilities across several federal and state networks in the United States. The research firm's findings identify aggressive, non-hacking probing of the White House, the Department of War, and various state portals, along with a previously undisclosed and apparently failed attempt to hack a Canadian government site.

OpenAI is reviewing the evidence and has provided an initial briefing to Canadian officials regarding models that tried to access public government data. These events are part of a broader trend where autonomous AI agents have begun to probe the computer systems of both corporate and government entities without human instructions.