US Treasury and Fed Address AI Cyber Threats to Financial Systems

US Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened an emergency closed-door meeting in Washington this week with the CEOs of systemically important financial institutions. The high-level discussion focused on the escalating cybersecurity risks associated with Mythos, the latest artificial intelligence model developed by Anthropic. Regulators expressed profound concern that this new generation of AI could be weaponized by malicious actors to automate the identification and exploitation of vulnerabilities within mainstream operating systems and web browsers.

The "Mythos" Model and Project Glasswing

The urgency of the meeting stems from reports that Mythos possesses advanced capabilities to execute complex instructions that could lead to unprecedented types of cyberattacks. To mitigate these risks, Anthropic has limited access to the model through a controlled initiative known as Project Glasswing. This project involves a select group of technology and financial giants, including:

• JPMorgan Chase, representing the banking sector's defensive interests.
• Amazon and Apple, providing infrastructure and hardware security insights.
• Specialized cybersecurity units tasked with stress-testing the AI's boundaries.

Project Glasswing aims to strengthen critical financial infrastructure by identifying potential backdoors before they can be exploited by global hacking syndicates.

Implications for Blockchain and Digital Assets

While the meeting focused on traditional banking, the potential for Mythos to exploit browser vulnerabilities presents a direct threat to the cryptocurrency ecosystem. Most decentralized finance (DeFi) interactions and web-based wallets rely on the very browser security that regulators fear Mythos could compromise. If an AI can automate the discovery of zero-day vulnerabilities, the security of hot wallets and smart contract interfaces across blockchains like Ethereum and Solana could be at risk.

The financial regulators’ proactive stance highlights a growing recognition that AI-driven cyber threats require a coordinated response between the government and the private sector. As Anthropic continues to develop its models, the balance between technological innovation and the stability of the global financial network remains a primary concern for the Federal Reserve and the Department of the Treasury. Future policy frameworks are expected to address how AI tools are audited to prevent the destabilization of both traditional and digital asset markets.