The $1.2 trillion-valued San Francisco-based artificial intelligence giant Anthropic is again publicly contracting with the second Trump administration, according to a newly public summary of a $1 delivery order between the Federal Communications Commission (FCC) and the Reston-based technology seller Carahsoft Technology Corp.
Actioned on July 8 with a start date of July 6, the official summary of the 85-day contract is “Anthropic Claude for Government Federal - FedRAMP High Compliant AI Cloud Service Subscription for FY26-Pilot,” in reference to a classification within the Federal Risk and Authorization Management Program (FedRAMP) suitable for sensitive but unclassified information. The new delivery order effectively recreates a previously cancelled pilot program which was initiated on February 23 and cancelled on March 11, under FCC administrator Brendan Carr.
One week prior to the first pilot program’s cancellation, Carr told CNBC that Anthropic had “made mistakes” in its dealings with the Pentagon, in which the company demanded assurances that the U.S. military would not use its products for either mass domestic surveillance or killing without human approval. President Trump responded in late February by demanding that all U.S. Government agencies “immediately cease” all usage of Anthropic’s products, with a six-month carveout granted to the Pentagon.
What became an effective marketing campaign for Anthropic began as a revelation that the company’s products had supported the Trump administration’s rendition of Venezuelan president Nicolas Maduro in early January.
Claude was also one of the large language models leveraged, as part of Palantir’s Maven Smart System targeting software, to conduct what was reported to be more than 2,000 strikes on Iran in a 96-hour period beginning in late February. Controversy has centered on the U.S. military’s February 28 missile strike on Shajarah Tayyebeh elementary school, which killed more than 100 Iranian children.
Anthropic’s status as a U.S. federal contractor was officially reactivated on June 12, following reporting from Financial Times the previous week that the U.S. National Security Agency (NSA) was beginning to leverage Anthropic’s Mythos cybersecurity product for conducting cyber attacks. According to FT, Anthropic “installed about half a dozen staff within the NSA as so-called forward-deployed engineers to guide the use of the technology and customise models for specific applications.”
The Australian Signals Directorate, Australia’s rough analogue of the NSA, has further publicly disclosed two small contracts with Anthropic for training services and software licenses since May.
As exclusively reported by this publication, Anthropic’s relationship with the CIA and Australia’s Office of National Intelligence dates back to at least November 2023, in close collaboration with former Google CEO Eric Schmidt’s think-tank Special Competitive Studies Project (SCSP) and the Australian Strategic Policy Institute (ASPI). Schmidt was one of the earliest investors in Anthropic, and Google itself has considered investing up to $40 billion.
In recent months Anthropic hired the former head of the Pentagon’s think-tank, James H. Baker, as a “strategist-in-residence” and appointed the technology executive Teresa Carlson, who helped launched Amazon Web Services’ Bahrain branch in 2019, as its new head of public sector.
(Disclosure: Carlson’s husband, C5 Capital CEO Andre Pienaar, hired the prominent defamation firm Clare Locke to threaten the author of this article with a lawsuit for referencing his previous involvement in a major Pentagon cloud procurement scandal.)