Concerns over Mythos’ AI capabilities have prompted U.S. regulators to organize an emergency meeting, aiming to evaluate the potential implications of Anthropic’s sophisticated artificial intelligence model on financial institutions.
The discussion took place Tuesday with Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell leading a session with executives from major Wall Street banks. Bloomberg reported that the meeting focused on cybersecurity risks associated with Mythos, based on information from sources familiar with the proceedings.
Chief executives from systemically important financial institutions, including Citigroup Inc., Morgan Stanley, Bank of America Corp., Wells Fargo & Co., and Goldman Sachs Group Inc., were in attendance. The systemic importance of these banks means that any operational disruptions could have far-reaching global impacts.
Developed by Anthropic, Mythos is an advanced AI model designed for cybersecurity purposes. Its primary function is to detect software vulnerabilities and bugs, yet it also has the potential to create sophisticated exploits when prompted. This marks a shift in regulatory perspectives on AI risks, now viewed not just as technological challenges but as possible triggers for systemic events.
The concerns extend into the cryptocurrency sector, where experts express unease over Mythos’ ability to find and exploit zero-day vulnerabilities at minimal cost, posing threats to decentralized finance infrastructure.
Anthropic has cautiously released Mythos through “Project Glasswing,” limiting access to a select group of large technology and financial firms. Prior to its launch, Anthropic consulted with U.S. officials regarding the model’s defensive and offensive capabilities. Additionally, the company is engaged in legal proceedings against the Pentagon over a designation as a supply-chain risk, which Anthropic contests in court.