Bessent, Powell Summon Bank CEOs Over Anthropic’s New AI Model  

Treasury Secretary Scott Bessent and Fed Chair Jerome Powell called Wall Street’s most powerful executives to Treasury headquarters in Washington on Tuesday, according to Bloomberg News. The gathering was unannounced and arranged on short notice. The concern at the center of it: Anthropic’s newly released AI model, called Mythos, and what it could mean for the security of the global financial system.

Regulators believe Mythos represents a new kind of threat. According to Bloomberg, the model can identify and exploit weaknesses in major operating systems and web browsers when directed to do so by a user. In the wrong hands — say, a criminal hacker or a hostile foreign actor — that capability could be devastating for banks and the broader financial system. All the banks summoned to Tuesday’s meeting are classified as systemically important, meaning regulators consider their stability critical to the global economy.

The CEOs who attended included Citigroup’s Jane Fraser, Morgan Stanley’s Ted Pick, Bank of America’s Brian Moynihan, Wells Fargo’s Charlie Scharf, and Goldman Sachs’ David Solomon. JPMorgan’s Jamie Dimon was unable to make it, Bloomberg reported. Spokespeople for the banks declined to comment.

Anthropic has not released Mythos to the general public. Instead, the company is rolling it out slowly through a program called “Project Glasswing,” which includes a handful of major technology and finance firms, among them Amazon, Apple, and JPMorgan Chase. The idea is to give those companies a head start on securing critical systems before similar AI models become more widely available.

Anthropic told Bloomberg it had been in discussions with US officials about Mythos and what it described as its “offensive and defensive cyber capabilities” before the model’s recent release. The company did not immediately respond to a request for comment on the emergency bank meeting.

Related: Anthropic Joins Forces With Amazon, Microsoft, Apple on Cyber Defense AI

Regulators’ caution mirrors Anthropic’s own careful approach to the rollout. But it also signals a broader shift in how Washington is thinking about AI. Powerful models are no longer just a business story. They are becoming a national security and financial stability issue fast.

The emergency meeting was not the only sign of turbulence surrounding Anthropic this week.

The San Francisco-based company is also fighting a separate legal battle with the Trump administration. The Pentagon designated Anthropic a supply-chain risk, a label the company has aggressively contested in court. Earlier this week, a federal appeals court declined to pause that designation while the legal fight plays out.

Anthropic is also navigating a rivalry with OpenAI, which sent a note to investors this week claiming it has a significant computing advantage over its competitor. OpenAI said it had 1.9 gigawatts of computing capacity available in 2025 — triple the year before — compared to an estimated 1.4 gigawatts for Anthropic. OpenAI used that gap to argue Anthropic’s limited rollout of Mythos may have been driven less by caution and more by computing constraints.

Anthropic pushed back on that framing. The company recently announced a major partnership with Broadcom and Google to access about 3.5 gigawatts of computing power starting in 2027, and has committed to spending $50 billion on US data centers.

For now, regulators, banks, and the public are watching closely. The Mythos rollout and the legal fight with the Pentagon are both moving quickly. More developments are expected on both fronts.