Anthropic Filing Reveals Pentagon Emailed Day After Blacklist Saying Sides Were "Very Close" — Calls DOD Claim of Approval Demand "A Central Falsehood"timeline_event

ai-safetymilitary-aiautonomous-weaponsanthropic-blacklist
2026-03-20 · 1 min read · Edit on Pyrite

type: timeline_event

On March 20, 2026, Anthropic filed sworn declarations in the Northern District of California that fundamentally contradicted the Pentagon's public narrative about why the company had been blacklisted. The filing revealed that on March 4 — the day after the supply chain risk designation was announced — Under Secretary Emil Michael had emailed Anthropic CEO Dario Amodei stating that the two sides were "very close" on the key issues of autonomous weapons policy and surveillance guardrails. The email directly undermined the Pentagon's claims that negotiations had broken down due to irreconcilable disagreements.

The declarations, submitted by Anthropic executives Sarah Heck and Thiyagu Ramasamy, provided a detailed timeline of the negotiations leading up to the blacklist. They showed that substantive progress had been made on nearly all contested points, and that the remaining gaps were narrow and potentially bridgeable. The filing specifically rebutted the DOD's central claim — made in its March 18 response — that Anthropic had demanded an "approval role" over military operations, calling this characterization "a central falsehood." Anthropic's declarations stated that the company had never sought veto power over military decisions but had instead proposed contractual limitations on specific use cases, similar to restrictions already present in contracts with other defense technology providers.

The contradiction between Michael's private email acknowledging the sides were "very close" and the administration's public posture that Anthropic was an unreasonable actor operating in "bad faith" raised pointed questions about the true motivation behind the blacklist. Legal observers noted that if the designation had been issued while the parties were near agreement, it suggested the decision was driven by political considerations — possibly a desire to make an example of a company that had publicly advocated for AI safety — rather than genuine supply chain security concerns. The filing was expected to feature prominently in the March 24 preliminary injunction hearing, where Anthropic would argue that the designation was arbitrary, retaliatory, and exceeded the government's statutory authority.