Bottom line: Anthropic CEO Dario Amodei met Friday with White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent in what the White House called “productive and constructive” talks about access to Mythos, the frontier AI model capable of finding thousands of zero-day vulnerabilities. The meeting marks a thaw in the standoff that began when the Pentagon blacklisted Anthropic for refusing to lift security restrictions, although any deal would likely exclude the Defense Department and route access to Mythos through civilian agencies.
Anthropic CEO Dario Amodei entered the West Wing on Friday to meet with White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent. The White House described the conversation as “introductory, productive and constructive,” and said the three discussed “opportunities for collaboration, as well as shared approaches and protocols to address the challenges associated with scaling this technology.” President Trump later told reporters that he had “no idea” the meeting had taken place.
The meeting is the most significant step toward resolving a standoff that has left one of the world’s largest artificial intelligence companies blacklisted by its own government, as that same government fights to gain access to its most powerful model. If the two sides reach a deal, it will likely exclude the Pentagon entirely, diverting access to Mythos through civilian agencies that are not parties to the original dispute.
How do we get here?
The conflict began in late February when Defense Secretary Pete Hegseth demanded that Anthropic grant the Pentagon unrestricted access to its AI models for “all lawful purposes,” including autonomous weapons systems and domestic surveillance. Amodei refused. He has said publicly that Anthropic wants to work with the military, but that AI models are not yet reliable enough for autonomous weapons and that US law has fallen short of protecting Americans around the use of AI in mass surveillance. Hegseth’s response was to designate Anthropic as a national security supply chain risk, a classification previously reserved for companies associated with foreign adversaries, effectively blacklisting it from all government contracts.
Anthropic sued the Trump administration in early March, filing two federal lawsuits alleging unlawful retaliation. A federal judge initially blocked the blacklisting, but an appeals court overturned that decision on April 8. Anthropic is now excluded from Department of Defense contracts, but can still work with other government agencies. After the court ruling went against it, Anthropic hired Trumpworld consultants to facilitate a political resolution, and Axios reported that Friday’s meeting was designed to pave the way for a deal.
The 💜 of EU technology
The latest rumors from the EU tech scene, a story from our wise founder Boris and some questionable AI art. It’s free, every week, in your inbox. Register now!
The paradox that brought Amodei to the White House is that Anthropic announced Mythos on April 7, ten days after it fell out of favor, and the model turned out to be something the government couldn’t ignore.
What can Myths do?
Mythos is a general-purpose AI model that, during testing, proved capable of identifying and exploiting thousands of previously unknown zero-day vulnerabilities across all major operating systems and web browsers. It found flaws that had survived decades of human safety review. When ordered to develop working exploits, it succeeded on the first try in more than 83% of cases. It is the first AI model to complete a 32-step corporate network attack simulation from start to finish. The UK’s AI Security Institute assessed it as “substantially more capable of committing cybercrime than any model previously evaluated.” JPMorgan Chase CEO Jamie Dimon said publicly that it “reveals many more vulnerabilities” to cyberattacks. The Council on Foreign Relations called it “a turning point for AI and global security.”
Anthropic decided not to publish Mythos. Instead, it created Project Glasswing, an access-controlled program that provides the blueprint to approximately 40 vetted organizations, including Amazon Web Services, Apple, Google, Microsoft, Nvidia and JPMorgan Chase, to find and fix vulnerabilities before they can be exploited. The company committed up to $100 million in Mythos usage credits and $4 million to open source security organizations. The decision to restrain rather than release is a direct application of the security principles that put Anthropic in conflict with the Pentagon in the first place.
What each side wants
The Treasury Department is looking to Mythos to look for vulnerabilities in its own systems. Parts of the intelligence community and the Cybersecurity and Infrastructure Security Agency are already testing it. The White House Office of Management and Budget is putting safeguards in place to allow federal agencies to use a controlled version. Bessent’s presence at Friday’s meeting indicates that the economic and financial security arguments for access to Mythos have reached the highest levels of the administration.
Anthropic needs the blacklist resolved. Not because it needs Pentagon revenue; he annualized company income has reached $30 billion, has attracted bids from investors at a valuation of $800 billion, and is exploring an initial public offering (IPO). But supply chain risk designation damages your business credibility and creates uncertainty for all government-adjacent customers. What Amodei wants is a resolution that restores his company’s reputation without giving up the safety commitments that sparked the dispute.
The general lines of a commitment are visible. Anthropic would provide access to Mythos for defensive cybersecurity purposes through civilian agencies. Management would remove or reduce the supply chain risk designation. The Pentagon would remain excluded unless a separate process can be agreed upon to review specific military use cases. Both sides have incentives: Anthropo because the blacklist is commercially harmful, and the White House because the technology is too valuable to give up.
The pressure from outside
The diplomatic dimension adds urgency. Anthropic plans to provide Mythos for selection British banks in a matter of days and is quadrupling its London office to 800 employees. Bank of England Governor Andrew Bailey named Mythos as a cybersecurity risk in a speech at Columbia University on April 15, and the Bank’s Cross-Market Operational Resilience Group is convening an emergency briefing with the chief executives of the UK’s eight largest banks and representatives from the Treasury, the FCA and the National Cyber Security Centre. Canadian Finance Minister François-Philippe Champagne described Mythos as an “unknown unknown” at IMF meetings.
The result is a situation in which America’s closest allies may have access to a critical national security tool before the U.S. government. That geopolitical reality It gives the White House an incentive to resolve the dispute that transcends the original disagreement over security barriers. Bessent, whose Treasury Department is one of the agencies most eager for access to Mythos, presumably made this point clear at Friday’s meeting.
what friday means
The word “introductory” in the White House reading is carefully chosen. He points out that Wiles and Bessent are opening a channel, not closing a deal. The litigation remains active. The appeals court ruling still stands. Hegseth has not retired his position. But the fact that the White House Chief of Staff and the Treasury Secretary sat down with the CEO of a company the Pentagon has blacklisted and described the conversation as productive represents a change in the administration’s stance that would have been difficult to imagine six weeks ago.
Amodei built the most capable cybersecurity tool in existence as a byproduct of building a general-purpose AI model, then restricted its release for security reasons, then was punished by the government for upholding those same security principles, and is now being courted by that government because the tool cannot be replicated or replaced. That sequence plays out not in a congressional hearing or a regulatory proceeding, but in a West Wing room where the most powerful chief of staff in a generation, the Treasury secretary, and the CEO of an artificial intelligence company are trying to find a formula that satisfies national security, commercial reality, and the security principles that started the whole fight. On Friday he did not produce that formula. But it established that everyone in the room wants one.
