AI and Financial System Risk

Federal Reserve Chair Jerome Powell and Treasury Secretary Scott Bessent held a closed‑door meeting this week at the Treasury Department in Washington, D.C., with top U.S. bank CEOs to discuss cybersecurity risks from Anthropic’s latest AI model, Mythos, according to sources cited by CBS News and earlier reporting by Bloomberg. JPMorgan Chase CEO Jamie Dimon was invited but did not attend. The session followed Anthropic’s announcement Tuesday of Project Glasswing, a partnership with Amazon, Apple, Nvidia and others to use Mythos defensively after the system exposed serious vulnerabilities in major operating systems and web browsers. Anthropic has decided not to widely release Mythos, warning that its offensive capabilities could soon proliferate and that the resulting economic, public‑safety and national‑security fallout could be severe if misused. The meeting underscores that U.S. financial regulators now see advanced AI models themselves as a potential systemic cyber threat, building on the Biden administration’s 2023 designation of AI as a possible risk to financial stability and feeding online debate over whether Washington is moving fast enough to regulate high‑end AI in critical infrastructure.