IC Technology & Surveillance — 2026-05-09

Pentagon Signs AI Deals with Eight Tech Companies While Maintaining Anthropic Freeze

The Washington Post reported on May 1 that the Pentagon signed classified AI agreements with Microsoft, Amazon, Nvidia, and Reflection AI. The new entrants join OpenAI, xAI, and Google as vendors cleared for deployment at Impact Levels 6 and 7 under terms permitting "any lawful use." Anthropic remains the only US-headquartered firm carrying a Pentagon supply chain risk designation and is suing over a cancelled $200 million contract, after CEO Dario Amodei publicly refused the "any lawful use" framing, per ResultSense. ResultSense, citing Axios, reported that White House officials are seeking a path to restore the relationship, and that Anthropic's Claude coding model has remained in active use by US government security organizations throughout the dispute.

Analysis
Expanding to eight vendors directly reduces the Pentagon's dependency on any single vendor's policy stance, weakening Anthropic's negotiating position even as unnamed White House officials, per Axios via ResultSense, seek a reconciliation path. Active litigation over the cancelled $200M contract and neither side's concession on "any lawful use" language make a bilateral climb-down politically difficult for both parties. A lift of the contracting freeze within 90 days of May 9, 2026 is unlikely. The White House reconciliation effort may be primarily political cover, with both parties content to let Anthropic's informal operational presence, with Claude coding tools reportedly still in use by US security organizations, substitute for a formal contract.

2 sources

Pentagon broadens AI defence suppliers as Anthropic dispute drags on - ResultSense
Pentagon will never again rely on a single AI provider, official says - Government Executive