Treasury Secretary and Fed Chair warn big banks on Anthropic Mythos AI risks
Updated
Updated · Bloomberg · Apr 25
Treasury Secretary and Fed Chair warn big banks on Anthropic Mythos AI risks
9 articles · Updated · Bloomberg · Apr 25
Treasury Secretary Scott Bessent and Fed Chair Jay Powell summoned major US bank leaders to Washington to address risks from Anthropic’s new Mythos AI system.
Experts highlight that Mythos operates at unprecedented speed, raising cybersecurity concerns and exposing the financial sector to challenges regulators have not previously faced.
The warnings come as the IMF addresses global economic instability linked to the US-Iran conflict, underscoring the growing importance of robust financial oversight during periods of technological and geopolitical disruption.
How does the financial system survive when facing simultaneous AI and geopolitical crises?
Can new financial regulations possibly keep pace with AI that evolves at 'hyper speed'?
Anthropic abandoned a key AI safety pledge. Can tech giants be trusted to self-regulate world-altering technology?
When AI can autonomously hack financial systems in hours, are traditional cybersecurity defenses now obsolete?
As war disrupts the AI supply chain, is the global tech investment boom already facing its end?
With data centers now being military targets, how can the global AI infrastructure be secured from physical attack?
Project Glasswing and the Race to Defend Against AI-Accelerated Cyber Exploits
Overview
In early April 2026, Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell urgently met with major U.S. bank CEOs to warn about the severe cybersecurity risks posed by Anthropic's Claude Mythos Preview AI. This advanced AI autonomously discovers and exploits critical software vulnerabilities at unprecedented speed, revealing a 99% unpatched rate and threatening financial stability. In response, Anthropic launched Project Glasswing, granting controlled Mythos access to select organizations for defensive use, though this has caused challenges like maintainer overload. The U.S. meeting spurred international regulatory actions, highlighting a global need for coordinated AI-native cybersecurity strategies to manage this rapidly evolving threat.