Five Systemically Important Banks Testing Anthropic’s Mythos AI Cyber Model
US Treasury Secretary Scott Bessent and Fed Chair Jerome Powell convened top banks including JPMorgan, Goldman Sachs, Citigroup, Bank of America and Morgan Stanley to test Anthropic’s Mythos AI model for cyber vulnerabilities. Anthropic’s Project Glasswing rollout includes Amazon among the initial dozen firms with model access.
1. Regulatory Push for AI Cybersecurity Testing
On April 7, US Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell summoned executives from systemically important banks to Treasury headquarters for a briefing on cyber threats. Participating institutions—including JPMorgan, Goldman Sachs, Citigroup, Bank of America and Morgan Stanley—were urged to deploy Anthropic’s Mythos model internally to scan for vulnerabilities within their own systems.
2. Anthropic’s Limited Rollout under Project Glasswing
Anthropic has confined Mythos access to a few dozen companies under “Project Glasswing,” focusing on securing critical infrastructure before broader release. Early adopters include JPMorgan, Amazon and Apple, with Mythos demonstrating capabilities to identify—and in tests potentially exploit—vulnerabilities such as unauthorized web browser data access.