The Federal Reserve raised concerns regarding the potential risks associated with Anthropic’s Mythos AI model and had to operate for several months without it. This incident occurs within a broader context where U.S. authorities are grappling with the dual-use cyber capabilities of advanced AI systems, emphasizing the need for vigilant management of technologies that could threaten critical infrastructure. Additionally, recent export-control directives have limited access to certain leading AI models for foreign nationals, reflecting the increasing regulatory scrutiny surrounding powerful AI technologies.

Fed: The Federal Reserve serves as the central bank of the United States, overseeing monetary policy, financial stability, and supervision of the banking system. In April 2026, Fed Chair Jerome Powell joined Treasury Secretary Scott Bessent in convening major bank CEOs to discuss emerging risks. The Fed has continued highlighting how advanced AI capabilities, such as those demonstrated by certain models, can rapidly alter cybersecurity landscapes for financial institutions.
Anthropic: Anthropic develops frontier AI systems focused on safety and capability, including the Claude family of models. It created Mythos, a general-purpose model noted for its ability to identify and exploit software vulnerabilities at scale. The company has faced U.S. government directives limiting access to advanced versions of the model due to dual-use concerns.

Security: Government and industry discussions continue around the dual-use cyber capabilities of leading AI systems and their implications for critical infrastructure.
Regulation: U.S. authorities have issued export-control directives requiring suspension of access to certain frontier AI models for foreign nationals.