The Trump administration is trying to get its hands on a powerful new artificial intelligence system built by the same company it has moved to blacklist across the federal government.
White House Chief of Staff Susie Wiles met with Anthropic CEO Dario Amodei to discuss what the Associated Press described as an AI model it called “Mythos,” focusing on its national-security and economic implications. No product by that name appears in Anthropic’s public documentation, and the company has not confirmed the designation. The sit-down took place even as a separate administration directive ordered federal agencies to stop using Anthropic’s technology and the Pentagon branded the San Francisco-based AI lab a supply chain risk.
That collision, a ban enforced by one arm of the government and a courtship led by another, points to a deep internal split over how Washington should handle one of the most advanced AI developers in the country.
A ban, a blacklist, and a West Wing meeting
The sequence of events has moved fast. According to AP reporting, the administration instructed agencies to halt their use of Anthropic systems, effectively cutting the company off from government work. The Pentagon went further, formally designating Anthropic a supply chain risk, a label that can bar a firm from defense contracts and trigger cascading restrictions across federal procurement.
U.S. Senator Ed Markey, a Massachusetts Democrat, responded by demanding immediate congressional action to reverse the Pentagon’s designation. Markey’s office called the move retaliatory, tying it to a dispute over the safety guardrails Anthropic had insisted on for its products. No press release link, specific date, or direct quote from the senator’s office has been made available for independent verification. That framing casts the blacklisting as a political punishment rather than a technical security judgment, though no internal Pentagon memos justifying the decision have surfaced publicly.
A federal judge then issued a preliminary injunction temporarily blocking the supply chain label. The ruling came after an emergency hearing and referenced a related case in its reasoning. While the injunction does not settle the dispute permanently, it freezes the designation during litigation, a signal that the court found either a likelihood Anthropic could prevail on the merits or a risk of irreparable harm if the label stood.
Against that backdrop, Wiles and Amodei sat down in the West Wing. The meeting’s focus on the model the AP called “Mythos,” rather than on resolving the ban, suggests senior officials see the system as a distinct asset worth pursuing regardless of the broader crackdown on the company.
Contradictory signals from inside the administration
Separate reporting indicated that White House officials are exploring ways to route the model to federal agencies, which would amount to a direct reversal of the administration’s own directive. No official readout or on-the-record statement has confirmed that effort, and the White House has not publicly explained how it squares with the order to cut Anthropic off. The sourcing for this claim has not been tied to a specific, publicly accessible article.
The most straightforward explanation is that different factions hold different views. National-security officials may see the system as too capable to leave on the shelf, while procurement and compliance teams are enforcing the supply chain designation as written. But that interpretation is analytical. No administration official has acknowledged the contradiction on the record.
Anthropic has not released detailed public documentation about what sets the reported model apart from its existing Claude family of products. Without specifics on the system’s capabilities, benchmarks, or intended deployment, it is hard to assess why the White House would treat it differently from the Anthropic products already banned, or why defense officials might view it as uniquely risky.
What agencies face right now
For federal offices that had been using Anthropic tools before the ban, the practical situation is stuck in limbo. The court injunction prevents the supply chain label from taking full effect, but it does not automatically restore previous contracts or guarantee future access. Procurement officers are navigating a landscape where one part of the government has questioned Anthropic’s reliability as a supplier while another has signaled interest in deeper collaboration.
The Pentagon’s role in the dispute, detailed in separate AP reporting, adds another layer of complexity. Defense Secretary Pete Hegseth’s department initiated the supply chain designation, and any reversal would likely require either a policy decision from within DOD or a court order with broader scope than the current injunction.
Meanwhile, rival AI firms stand to benefit from Anthropic’s exclusion. Companies like OpenAI and Google, which have aggressively pursued federal contracts, could absorb work that would otherwise go to Anthropic, reshaping the competitive landscape for government AI procurement for years.
Why the courts, Congress, and the West Wing are pulling in different directions
Senator Markey’s characterization of the Pentagon’s move as retaliation is a political claim, not a judicial finding. It carries weight because it came through official channels from a sitting senator, but it reflects one lawmaker’s reading of events. The federal judge’s decision to block the designation on procedural grounds lends indirect support to the idea that the Pentagon’s process was flawed, yet a procedural deficiency and political retaliation are not the same thing. A court can find that an agency skipped required steps without endorsing any theory about motive.
For the White House, the episode exposes the difficulty of managing frontier AI within existing legal and bureaucratic frameworks. A system like the one the AP called “Mythos” can look like a security asset, a commercial product, and a regulatory problem all at once. When different parts of the government emphasize different facets of that identity, policy fractures, producing exactly the mixed signals now on display: a high-level meeting in the West Wing, a blacklist inside the Pentagon, and a court order pressing pause on the most aggressive action.
Until more internal documents, technical disclosures, or official statements become public, Washington has not settled on a coherent Anthropic policy. The Wiles-Amodei meeting shows that senior officials want to understand the model. The Pentagon designation and the reported ban show deep institutional resistance. The injunction shows the courts are willing to check how far the executive branch can go in sidelining a major American AI company. All three threads will shape what comes next, but as of May 2026, they coexist in unresolved tension.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
