Anthropic CEO Dario Amodei has held renewed discussions with senior U.S. defense officials as part of a renewed push to finalize terms for a contract allowing the Department of Defense to use the startup's artificial intelligence models. People familiar with the matter said the conversations are aimed at overcoming sticking points that derailed earlier talks.
Those familiar with the talks said Amodei has been in contact with Emil Michael, the under-secretary of defense for research and engineering, in an effort to bridge differences and secure agreement on the contract. The discussions come after previous negotiations between Anthropic and the Pentagon collapsed.
According to those close to the negotiations, the earlier breakdown prompted a sharp response from defense leadership. Defense secretary Pete Hegseth had threatened to label the AI firm a supply chain risk in the wake of the failed talks.
Central to the dispute were contractual provisions Anthropic regarded as critical to ensuring its technology would not be employed for mass domestic surveillance or used to enable lethal autonomous weapons. In an internal memo shared with staff, Amodei said the disagreements with the Pentagon revolved around the "analysis of bulk acquired data," language the company viewed with suspicion.
Amodei also told employees that the company believes it was sidelined in part because it did not offer what he described as "dictator-style" praise for U.S. President Donald Trump, according to reports cited by sources.
Anthropic is backed by several major U.S. technology firms, including Amazon and Alphabet. The company had signed a $200 million agreement with the Pentagon in July 2025 under which its AI models were slated to be the first used with the defense department's classified data.
After negotiations between Anthropic and the Pentagon broke down, rival OpenAI announced a contract with the Pentagon. Sources also reported that OpenAI is pursuing a separate contract with the North Atlantic Treaty Organization.
Key points
- Anthropic CEO Dario Amodei has resumed talks with senior Defense Department officials, including Emil Michael, to try to finalize a contract for the use of Anthropic's AI models.
- Negotiations previously collapsed amid disagreements over safeguards to prevent domestic surveillance and lethal autonomous weapons, and concerns over provisions described as "analysis of bulk acquired data."
- The dispute affected the defense and AI sectors, with the Pentagon moving to contracts with rival firms after talks with Anthropic broke down.
Risks and uncertainties
- Continued disagreement over data-analysis and usage terms could prevent a finalized contract, leaving defense procurement plans for AI models uncertain - impacting defense technology acquisition.
- The potential designation of an AI vendor as a supply chain risk introduces regulatory and reputational uncertainty for firms working with defense suppliers - affecting companies in the defense and technology sectors.
- The competition among AI firms for defense and NATO contracts adds uncertainty to market allocation and partnership outcomes in the AI and defense contracting spaces.