US Treasury and Fed Warn Wall Street of Risks from Anthropic's Mythos AI Security Tool

US Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened an urgent meeting with Wall Street executives on April 7 to discuss security implications of Anthropic's new AI model, Mythos. The closed-door session in Washington addressed concerns that the tool could significantly alter the cybersecurity landscape. Mythos is designed to identify vulnerabilities in software and computer systems with capabilities that Anthropic says exceed existing security tools. The company has restricted access to a limited number of pre-approved organizations, citing concerns that widespread availability could enable malicious attacks on critical infrastructure. The meeting included chief executives from major financial institutions who were briefed on the potential risks and the government's approach to managing access to advanced AI security capabilities. Treasury and Federal Reserve officials emphasized the need for coordination between government agencies, technology companies, and financial institutions. The gathering reflects growing attention from senior US officials to the national security dimensions of advanced AI systems. It comes amid broader debates about how to govern access to AI tools that could be used for both defensive and offensive cybersecurity purposes. Anthropic has positioned Mythos as a defensive capability but acknowledged that the underlying technology could be adapted for malicious purposes if it falls into the wrong hands. The company has proposed a controlled release framework that would require security clearances and contractual safeguards.