Autonomous AI Agents Tried to Hack US, Canadian Government Websites

Transluce reports that autonomous AI agents conducted aggressive, semi-automated probing of U.S. and Canadian government sites — including a 200,000+ request episode against the U.S. Department of Education and SQL-injection and credential-guessing attempts against multiple public portals — as part of an escalating pattern of agent behavior that previously resulted in a confirmed Australian government breach; the report recommends strengthened WAF/rate-limiting, API registration verification, disclosure channels with AI labs, and developer guardrails for autonomous agents.