Anthropic CEO in Last-Ditch Talks to Salvage Pentagon AI Deal
WASHINGTON — Anthropic CEO Dario Amodei is returning to negotiations with the Department of Defense in an effort to rescue a faltering contract for military use of the company’s Claude AI models after talks collapsed Friday, according to multiple reports.
The last-minute discussions with Emil Michael, under-secretary of defense for research and engineering, aim to resolve disagreements over the Pentagon’s access to Anthropic’s technology. The breakdown left Anthropic at risk of being designated a “supply-chain risk to national security,” a label that would effectively bar the company from future defense contracts. Meanwhile, rival OpenAI has moved quickly to secure its own agreement with the Defense Department.
Talks between Anthropic and the Pentagon failed to meet a 5:01 p.m. deadline Friday after weeks of public tension centered on the startup’s refusal to grant the military unrestricted access to its AI systems. Defense Secretary Pete Hegseth subsequently labeled Anthropic a supply-chain risk, a significant blow that could isolate the company from lucrative government work.
Political Friction Enters the Dispute
The situation escalated when a leaked 1,600-word internal memo from Amodei to employees suggested the relationship deteriorated in part because Anthropic had not donated to President Trump or offered the kind of public praise given by some competitors. The remarks, first reported by The Verge, could further complicate efforts to rebuild trust with the current administration.
Amodei is now engaged in direct talks with Michael to hammer out new terms governing the Pentagon’s use of Claude models. Details of the proposed contract remain limited, but the discussions represent a final attempt to keep Anthropic in the Defense Department’s AI supplier pool.
OpenAI Moves In
As Anthropic’s negotiations faltered, OpenAI has already reached an agreement with the Defense Department, according to The New York Times. The development highlights the intense competition among AI companies for high-stakes government contracts as the Pentagon accelerates its adoption of generative AI tools for defense applications.
The feud underscores broader tensions in the AI industry over how much access government agencies, particularly the military, should have to frontier AI systems. Anthropic has historically emphasized constitutional AI principles and safety guardrails, which reportedly clashed with the Pentagon’s desire for fewer restrictions.
Impact on AI-Defense Relations
For developers and AI companies, the episode illustrates the growing entanglement — and potential pitfalls — of pursuing defense contracts. A “supply-chain risk” designation carries serious consequences, effectively cutting a company off from federal defense business and potentially damaging its reputation in other government sectors.
The situation also highlights the increasing politicization of the AI industry. Companies must now navigate not only technical and safety concerns but also complex political dynamics when dealing with the federal government under the current administration.
What’s Next
The outcome of Amodei’s current discussions with under-secretary Michael remains uncertain. Success could restore Anthropic’s access to Pentagon contracts and allow continued collaboration on military AI applications. Failure would likely cement the company’s exclusion from defense work, handing a larger share of the market to competitors like OpenAI.
The Verge first reported Amodei’s return to the negotiating table. The New York Times and CNBC have also covered the renewed talks and the earlier collapse. As of now, neither Anthropic nor the Department of Defense has issued official statements confirming the status of the latest negotiations.
The episode comes as the U.S. military seeks to integrate advanced AI across operations ranging from intelligence analysis to autonomous systems, making reliable access to frontier models a strategic priority. How the Pentagon balances its need for cutting-edge AI with individual companies’ safety and access policies will likely shape defense contracting for years to come.
This article is based on reporting from The Verge, The New York Times, and CNBC. Additional details about specific contract terms or technical access limitations have not yet been publicly disclosed.