Trump Administration Pushes Banks to Test Anthropic's Powerful New Mythos Model—Despite Court Battle

An Unlikely Alliance Amid Legal Conflict

In a striking turn of events, Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell have summoned major bank executives to encourage them to test Anthropic's newly announced Mythos model for detecting security vulnerabilities—even as the Trump administration is actively suing the AI company in federal court. The move reveals a complex dynamic within the administration, where national security interests in AI-assisted cybersecurity appear to outweigh ongoing litigation over government access restrictions.

Anthopic unveiled Mythos this week, though the company immediately announced it would severely limit access to the model. The stated reason was counterintuitive: Mythos is so effective at uncovering security vulnerabilities that the company deemed it too risky for widespread distribution, despite the model not being specifically trained for cybersecurity purposes. Yet that hasn't stopped government officials from quietly encouraging financial institutions to begin testing it.

"Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell summoned bank executives for a meeting this week where they encouraged the executives to use Anthropic's new Mythos model to detect vulnerabilities."

The Banks Rushing to Test Mythos

While JPMorgan Chase was publicly listed as the sole initial partner organization with access to Mythos, reporting reveals a broader testing initiative across Wall Street's most powerful institutions. Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley are all reportedly running tests on the model, according to Bloomberg. The speed and breadth of this adoption suggests coordinated encouragement from Treasury and Federal Reserve leadership rather than organic interest from the banks themselves.

This development carries significant implications for financial system security. Banks have long struggled with detecting sophisticated cyber threats, and an AI model capable of identifying vulnerabilities that weren't specifically trained to do so represents a potential leap forward in defensive capabilities. The fact that multiple megabanks are simultaneously testing suggests the government views this as critical infrastructure protection.

The Elephant in the Room: The Courtroom Battle

What makes this situation particularly peculiar is the underlying legal conflict. Anthropic is currently embroiled in federal court with the Trump administration over the Department of Defense's designation of the company as a supply-chain risk. That designation emerged after negotiations between Anthropic and the government collapsed—negotiations that centered on the company's efforts to restrict how its AI models can be used by U.S. government agencies.

"Anthropic is currently battling the Trump administration in court over the Department of Defense's designation of Anthropic as a supply-chain risk."

The DOD's supply-chain risk label is a serious matter that can restrict business relationships and access to federal funding. Anthropic fought this designation because it stems from fundamental disagreements about government use cases—specifically, the company's resistance to enabling military or surveillance applications without strict safeguards. Yet here we are, with Treasury and the Fed effectively working to expand Anthropic's footprint in the financial sector while the same administration litigates against the company elsewhere.

This contradiction suggests either a lack of coordination within the Trump administration, or a pragmatic calculation that Mythos's cybersecurity capabilities are too valuable to ignore regardless of broader policy disputes with Anthropic.

International Regulatory Concerns

The Mythos rollout is drawing attention beyond U.S. borders. The Financial Times reports that U.K. financial regulators are already discussing risks posed by the model. This suggests that as banks adopt Mythos for vulnerability detection, international supervisors are beginning to consider whether such a powerful tool could concentrate risk or create new attack surfaces if compromised.

Regulatory caution here is warranted. A model so effective at finding vulnerabilities that its creators felt compelled to restrict access might also present operational or security risks if it's deployed across interconnected financial institutions without sufficient oversight. The fact that foreign regulators are watching closely indicates that this could become a coordination point for international financial stability discussions.

The Hype vs. Reality Question

Some observers have suggested that Anthropic's justification for limiting Mythos access—that it's too good at finding vulnerabilities—may be either marketing hype or a calculated enterprise sales strategy rather than a genuine safety concern. The argument goes that restricting supply artificially inflates demand and positions the model as uniquely powerful. Whether or not that skepticism has merit, the practical reality is that major financial institutions are getting access and beginning to integrate Mythos into their security operations.