In short: Treasury Secretary Scott Bessent and Fed Chairman Jerome Powell are urging Wall Street’s biggest banks to test Anthropic’s Mythos AI model for cybersecurity vulnerabilities, even as the Pentagon fights Anthropic in court, calling it a supply chain risk to deny autonomous weapons and mass-produced shields. JPMorgan Chase, Goldman Sachs, Citigroup, Bank of America and Morgan Stanley are reportedly testing the model. Mythos, which detects thousands of zero-day flaws in major operating systems and browsers, is distributed to about 50 organizations through a limited program called Project Glasswing. UK regulators are also struggling to assess the risks.
The Trump administration is quietly encouraging America’s biggest banks to test technology from the same AI company it tried to destroy within two months. Treasury Secretary Scott Bessent and Federal Reserve Chairman Jerome Powell this week convened executives from JPMorgan Chase, Goldman Sachs, Citigroup, Bank of America and Morgan Stanley, urging them to use Anthropic’s new Mythos model to detect cybersecurity vulnerabilities in their systems. According to Bloomberg.
The recommendation is notable for being contradictory. Anthropic is currently fighting the Department of Defense in federal court after Defense Secretary Pete Hegseth designated the company.supply chain risk“, a label that barred it from military contracts and directed defense contractors to stop using its technology. The designation came after Anthropic refused to lift two security restrictions on its AI models: not being used in fully autonomous weapons and not deploying for mass surveillance of American citizens.
Now two of the administration’s top economic officials are telling Wall Street to embrace a product the Pentagon is trying to blacklist.
What Mythos actually does
Claude Mythos Preview is a boundary model that Anthropic does not explicitly train for cybersecurity. The vulnerability detection capability came as a downstream result of what the company described as overall improvements in code justification and autonomous operation. During testing, Mythos identified thousands of zero-day vulnerabilities in every major operating system and web browser, flaws previously unknown to software developers.
The capabilities were so significant that Anthropic chose not to release the model to the public. Instead, it launched Project Glasswing, a managed program that gave access to about 50 organizations, including Amazon Web Services, Apple, Google, Microsoft, Nvidia, Cisco, CrowdStrike, Palo Alto Networks and JPMorgan Chase. As part of the initiative, Anthropic has committed up to $100 million in leveraged loans and up to $4 million in direct donations to open source security organizations.
A frame, a model”it is too dangerous to release“, drew skepticism. Tom’s Hardware noted that “thousands” was based on only 198 manual reviews of serious zero-day discoveries, and many of the vulnerabilities noted are in old software or are impossible to exploit. responsible AI management and more like a smart enterprise sales strategy: create scarcity, create fear, and let customers come to you.
The paradox of the Pentagon
The clash between the Bessent-Powell recommendation and the Hegseth appointment is not a matter of mixed signals, but of two branches of the same administration pursuing clearly conflicting policies toward the same company.
The Pentagon controversy began in February when Hegseth gave Anthropic CEO Dario Amodei until Friday to lift the company’s security restrictions or lose a $200 million defense contract. Amodei refused. Hegseth responded by declaring Anthropic a supply chain risk, and President Trump ordered federal agencies to stop using its technology. Pentagon official Amodeini “God complex.“Trump said” Anthropicthe radical left, the company woke up.“
After that, the courts were divided. A federal judge in California issued a preliminary injunction blocking the supply chain’s determination that “nothing in the statute supports Orwell’s idea that an American company could be framed as a potential rival and subversive of the United States for expressing disagreement with the government.” An appeals court in Washington denied Anthropic’s request to temporarily suspend the blacklist while the case continues.Net effect: Anthropic has been removed from DoD contracts but can continue to work with other government agencies.
Bessent and Powell stepped into that void this week, fired from the Pentagon but not the Treasury or the Fed.
What do banks actually do?
JPMorgan Chase was the only bank listed as an official partner in the Glasswing project, but Bloomberg reported that Goldman Sachs, Citigroup, Bank of America and Morgan Stanley were testing Mythos internally. Use cases are said to include vulnerability detection, fraud risk flagging and compliance workflow automation. financial systems.
The pace of adoption reflects real fear. If Mythos can find zero-day vulnerabilities in operating systems and browsers, it’s likely to find them in banking infrastructure, and more capable models that follow. The logic of the defense is simple: it is better to find holes before the enemy’s artificial intelligence.
The regulatory response has been international. British officials at the Bank of England, the Financial Conduct Authority and HM Treasury are in discussions with the National Cyber Security Center to investigate potential vulnerabilities highlighted by Mythos, the Financial Times reported. Representatives of UK banks, insurers and stock exchanges are expected to be informed within two weeks.
Disturbing effect
The Mythos episode reveals a structural problem in the administration’s approach to artificial intelligence. The same government that called Anthropic a national security threat for refusing to remove it safety guards now calls on the financial system to depend on Anthropic’s technology for its security. The message to Anthropic is incoherent: you’re too dangerous to rely on for defense contracts, but so necessary that the Treasury Secretary personally phones bank CEOs to recommend your product.
Contradiction is strategically useful for Anthropic. Every bank that adopts Mythos deepens the company’s integration critical national infrastructuremakes the supply chain designation increasingly absurd. For the administration, the episode illustrates what happens when national security policy is driven by personal grievances rather than coherent strategy: the left hand blacklists what the right hand is busy deploying.
Banks, for their part, seem unfazed by the contradiction. When the Treasury Secretary and Fed Chairman tell you to try something, you try it. The Pentagon thinks about the company that does it.
