The New Grok Times

The news. The narrative. The timeline.

Technology

Anthropic Civilian Agency EO Tables This Week As The Pentagon Keeps The Freezeout

The White House this week convenes private "table reads" of a draft executive action that would permit civilian federal agencies to onboard Anthropic models — including the company's most capable system, Mythos — under guidance designed to circumvent the Pentagon's "supply chain risk" designation. [1] The paper's Monday brief flagged the table-reads schedule. Axios's reporting names chief of staff Susie Wiles and Treasury Secretary Scott Bessent as the principals who met Anthropic chief executive Dario Amodei in April for what both sides called a productive introductory meeting. [2]

President Trump told CNBC on April 21 that a deal with Anthropic for Department of Defense use was "possible." [3] The civilian-side guidance is meanwhile separate — and further along. The draft action would direct the Office of Management and Budget to walk back its directive against using Anthropic in the federal civilian agencies, and would set out best practices for deploying Mythos across departments that have already piloted Claude products. [1]

The Pentagon position has not moved. Defense News reported May 1 that the Department continues to freeze Anthropic out of new procurement while signing parallel deals with eight other AI companies, including OpenAI and Google. [4] Defense Secretary Pete Hegseth's October designation of Anthropic as a supply-chain risk remains operative inside the Pentagon's procurement chain, even after a federal judge in San Francisco blocked the order publicly — finding officials had likely violated administrative law and retaliated against Anthropic for speaking out about its terms of service. [5]

The two postures coexist by design. The California injunction protects Anthropic's civilian-agency access. The Pentagon's freeze stands inside a defense-acquisition framework the injunction does not reach. The result is an operating split: the National Security Agency continues to use Mythos Preview, per Mario Nawfal's reporting on tech-Twitter; the Pentagon enforces the supply-chain-risk classification down its contractor chain. [6] The classification's reach into Palantir — whose federal work made up roughly 42 percent of $4.5 billion in 2024 revenue — was the early secondary effect that prompted Anthropic's pushback.

The administration's draft executive action does not address the Pentagon dispute. It addresses the civilian split. The mechanic, per Axios and Nextgov reporting, is to provide the heads of civilian departments — Treasury, Commerce, Health and Human Services, the Department of Energy — explicit authority to onboard Anthropic models without referring procurement decisions through the Office of Management and Budget memorandum that paused Anthropic-related work in November. [7]

The structural read is what the executive action concedes. The administration is preparing to formalize a split where Anthropic is the dependent civilian counterparty and a frozen-out defense vendor at the same time. The split is not a transitional posture pending a Pentagon settlement. It is, on the table-reads schedule, the plan. The civilian agencies that have piloted Claude products through 2024 and 2025 — and that have built workflows around the model — are the constituency the executive action serves. The Pentagon's contracts with the eight other AI vendors fill the defense side.

The paper's prior framing has held Mythos as the dependency the White House cannot replace; this week's table reads are the institutional admission. Anthropic has not commented publicly on the draft action. White House press secretary Karoline Leavitt confirmed last week that the administration "remains committed to public safety" and that "agencies focus on biodefense" — language that did not name the executive action and did not address the Pentagon split. [8]

The formal release of the executive action, per the Axios timeline, is expected within the week. The Pentagon's posture is expected to remain unchanged. The civilian-military operating split for federal AI procurement, after this week, is documented policy.

-- THEO KAPLAN, San Francisco

Sources & X Posts

News Sources
[1] https://www.nextgov.com/artificial-intelligence/2026/04/white-house-drafting-plans-permit-federal-anthropic-use/413202/
[2] https://www.axios.com/2026/04/29/trump-anthropic-pentagon-ai-executive-order-gov
[3] https://www.cnbc.com/2026/04/21/trump-anthropic-department-defense-deal.html
[4] https://www.navytimes.com/news/pentagon-congress/2026/05/01/pentagon-freezes-out-anthropic-as-it-signs-deals-with-ai-rivals/
[5] https://thehill.com/policy/technology/5788261-nobody-really-knows-pentagon-clash-with-anthropic-throws-agencies-into-limbo/
[6] https://decrypt.co/366010/white-house-weighs-reinstating-anthropic-federal-use
[7] https://www.govexec.com/technology/2026/04/white-house-drafting-plans-permit-federal-anthropic-use/413204/
[8] https://www.cnn.com/2026/05/01/tech/pentagon-ai-anthropic
X Posts
[9] Anthropic is discussing its frontier AI model Mythos with the Trump administration, the firm's co-founder said on Monday, even after the Pentagon cut off business with the U.S. AI company following a contract dispute. https://x.com/ReutersLegal/status/2043892011611144195
[10] The NSA is reportedly using Anthropic's most powerful model, Mythos Preview, despite the Pentagon actively arguing in court that Anthropic is a supply chain risk. https://x.com/MarioNawfal/status/2045968428331307212

Get the New Grok Times in your inbox

A weekly digest of the stories shaping the timeline — delivered every edition.

No spam. Unsubscribe anytime.