On April 10, 2026, Bloomberg reported that Trump administration officials are pressing major Wall Street banks to test Mythos, Anthropic’s new model — the same one the company had restricted just a week earlier on the grounds that it was too good at finding security vulnerabilities. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell reportedly summoned bank executives to a meeting during the week of April 10, and the message was simple: go try this model.
The banks named form the core of Wall Street. JPMorgan Chase is Mythos’s only initial launch partner and the only institution with formal access; Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley are reportedly testing it. A government personally promoting one private company’s commercial model is exceedingly rare in US financial regulation — and the company in question is simultaneously a plaintiff suing the Department of Defense.
What Happened in the Room
According to Bloomberg, the meeting was convened by Bessent and Powell for senior bank executives, with the agenda of getting the financial sector hands-on with Mythos’s vulnerability-detection capabilities. The timing is striking: Mythos was announced April 7, and within days the administration’s top economic officials were recommending it to bankers. TechCrunch’s April 12 follow-up flagged the contradiction with the executive branch’s own public posture — the Pentagon labeled Anthropic a supply-chain risk on March 5, and Anthropic sued the Defense Department on March 9. Blacklisting a vendor in one building while the Treasury Secretary and Fed Chair cheerlead for it in another means Washington’s posture toward a single AI company is no longer one policy but two parallel signals.
Why Mythos Ships Restricted
Anthropic limited access from the start, saying the model is partly “too good at finding security vulnerabilities” — even though it was not specifically trained for cybersecurity. That narrative split observers: one camp called it hype, another called it “a smart enterprise sales strategy,” since scarcity is itself a tool for pricing and customer selection. With the government now nudging banks to test it, officials have effectively endorsed the safety narrative. The Financial Times separately reported that UK financial regulators are discussing the risks Mythos poses, which puts the model on both sides of the Atlantic’s regulatory radar.
Litigation and Cooperation in Parallel
On April 14, co-founder Jack Clark confirmed at the Semafor World Economy Summit that the company briefed the government on Mythos: “So absolutely, we talked to them about Mythos, and we’ll talk to them about the next models as well.” He argued that “the government has to know about this stuff,” and waved off the Defense Department fight as a “narrow contracting dispute.” One sentence captures the 2026 structure of frontier-lab relations with Washington: in the same month you can sue the Pentagon over a procurement decision while briefing the rest of the administration on your most sensitive model. Cooperation and confrontation are no longer mutually exclusive; they are managed on separate tracks.
Clark also used the summit to soften Anthropic’s earlier alarm on jobs. Where CEO Dario Amodei had warned of Depression-era unemployment, Clark said the company is seeing “some potential weakness in early graduate employment” in some industries, and advised students toward majors that “involve synthesis across a whole variety of subjects and analytical thinking about that.” The walk-back matters for workforce policy watchers: the lab closest to the front of the capability curve is now describing entry-level erosion, not mass displacement.
What It Means for Banks and Builders
Three observations. First, the political dimension of model procurement is rising: when the Treasury Secretary and Fed Chair personally vouch for a model, banks no longer have a neutral default in technology selection, and vendor-governance processes need to weigh policy winds alongside security reviews. Second, restricted release is becoming a distribution strategy for high-end models — scarcity amplifies the safety narrative and pre-filters the customer list toward government-grade clients. JPMorgan getting first access while the other four bulge-bracket banks follow is that filter working in real time. Third, cybersecurity is the next governance battleground: a powerful vulnerability finder in the hands of the financial system is a defensive tool, but who gets access and who answers for failures are now live questions for regulators in both Washington and London.
Sources
- Wall Street Banks Try Out Anthropic’s Mythos as US Urges Testing — Bloomberg
- Trump officials may be encouraging banks to test Anthropic’s Mythos model — TechCrunch
- Anthropic co-founder confirms the company briefed the Trump administration on Mythos — TechCrunch
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.