Big bank CEOs (excluding Jamie Dimon) were urgently summoned by Fed Chair Jay Powell and Treasury Secretary Scott Bessant for a confidential meeting on April 7.
The meeting focused on Anthropic's AI tool "Mythos," which can detect extreme vulnerabilities in software and security systems, representing a new era of cyber risk.
Anthropic has acknowledged the severity and is conducting a limited rollout to a select group of ~40 companies for testing, with JPMorgan publicly confirmed as one.
The joint involvement of Powell and Bessant signals cross-government priority for financial stability and U.S. economic security, marking an unusual coordination.
Wider context includes adversarial states (Iran, China, Russia) developing similar cyber capabilities, escalating the threat landscape.
Banks are systemically important but often have outdated or "crunchy" technology (e.g., Citigroup's past "fat fingers" incident), increasing vulnerability.
The Federal Reserve plays a key supervisory role, with staff inside banks evaluating defenses and new tools like Mythos.
Potential for new regulations or formal rules for banks to address cyber risks, with chatter around best practices and requirements.
Risk that Mythos or similar tools could be replicated from open-source code, making the technology accessible to malicious actors worldwide.
The race between cyber attackers and defenders is intensifying, with AI tools rapidly advancing, requiring a top-down, serious approach from institutions.