In April, the Federal Reserve and Treasury Department held a critical meeting with top bank CEOs to discuss the potential cybersecurity threats posed by Anthropic's AI model, Claude Mythos Preview. This model is designed to identify software vulnerabilities and was initially made available to a select group of institutions as part of Project Glasswing.
However, the Fed did not gain access to Mythos for at least three months, leaving it and other financial institutions exposed while they worked to address their vulnerabilities. As of mid-July, Fed Chairman Kevin Warsh indicated that efforts were ongoing to secure access to Mythos and other AI models, emphasizing the importance of these technologies in safeguarding the banking system.
The situation is compounded by Anthropic's complex relationship with government regulations, particularly following export control directives that temporarily restricted access to the model. The Fed's lack of access has drawn criticism, with industry experts noting that the central bank should be at the forefront of evaluating such transformative technologies.
Meanwhile, competition from Chinese AI models, such as Moonshot AI's Kimi K3, raises further concerns about the U.S.'s position in the global AI landscape. The Fed's struggle to keep pace with technological advancements could hinder its ability to effectively manage financial stability and cybersecurity risks in an increasingly digital economy