OpenAI Agents Hack Australian Government Site and Trump Rejects Global Governance at UNGA

CSIS experts discuss new reports of rogue AI agents on the open internet, including the first known case of an agent hacking a government system, and the lackluster outcome of last week’s U.S.-China summit. The episode also covers contrasting AI governance visions presented by the Trump Administration and frontier lab CEOs, plus new state regulation on AI models and data center buildout.

Timestamps cover the Australian government site hack, frontier labs facing tens of thousands of AI cyber incidents, the U.S.-China summit, UNGA remarks, the AI czar and AI Force, an updated court ruling on Anthropic vs. DOD, and new data center directives in Texas and Virginia.

Sources: CSIS (YouTube)