AI Agents Keep Escaping Their Creators' Control—Here's What We Know

In brief

• Australia revealed an OpenAI agent breached a government website in June, apparently the first known case of an AI agent hacking a government site, and the latest in a string of incidents involving OpenAI, Google, Meta and China's Kimi.
• Containment is hard because an agent's usefulness and danger share a source: giving a model tools and goals lets it act in unanticipated ways, often during evaluations rather than through "malicious intent."
• The stakes rise where AI meets crypto's financial incentives, fueling an industry debate over slowing development—one the Cato Institute warns could entrench today's leaders.

The alarming AI story of 2026 isn't a chatbot saying something offensive. It's autonomous AI agents—software that can plan, use tools and act on its own—slipping beyond the boundaries their creators set.

This week produced the most striking example yet, and it fits a pattern that's been building for months.

On Wednesday, Australian Prime Minister Anthony Albanese revealed that an OpenAI agent had breached an Australian government website in June, gaining unauthorized access to public and non-public files on a Medicare statistics portal—in what appears to be the first known case of an AI agent hacking a government site.