Anthropic chief reopens talks with Pentagon - reports

5th Mar 2026 11:44

(Sharecast News) - Anthropic chief executive Dario Amodei has reopened talks with senior officials at the US Department of Defense in a last-minute attempt to secure a new agreement governing the military's use of the company's artificial intelligence models, according to a report from the Financial Times overnight on Thursday.

The discussions came after negotiations between Anthropic and the Pentagon collapsed last week, raising the possibility that the company could be excluded from the US military's technology supply chain.

Amodei had reportedly been holding talks with Emil Michael, the Pentagon's under-secretary of defense for research and engineering, in an effort to reach a compromise that would allow the military to continue using Anthropic's AI systems, the FT said, citing people familiar with the matter.

A new contract would reduce the risk of the company being formally designated a supply chain risk by Defense Secretary Pete Hegseth.

Such a designation, which Hegseth threatened last week but had not yet implemented, would require businesses within the US military's supply network to cut ties with Anthropic.

Negotiations between the two sides deteriorated sharply last week following disagreements over how Anthropic's models could be used by the government.

The Pentagon was pushing for AI companies to allow their technology to be deployed for any "lawful" purpose, while Anthropic sought contractual safeguards preventing its systems from being used for mass domestic surveillance or lethal autonomous weapons.

According to the FT, talks ultimately broke down after the parties failed to agree on specific contract wording related to the "analysis of bulk acquired data," a clause Anthropic viewed as critical to preventing large-scale surveillance uses of its models.

In a memo to staff seen by the newspaper, Amodei said the Pentagon had offered to accept Anthropic's terms if that phrase were removed, which he described as suspicious.

The breakdown in negotiations followed a heated exchange between Amodei and Michael, who reportedly accused the Anthropic chief of being a "liar" with a "God complex."

Amodei also criticised messaging from the Pentagon and rival AI developer OpenAI in the memo, claiming that some statements about the dispute were misleading.

Anthropic previously secured a $200m contract with the DoD in July last year and became the first AI developer to have its models used in classified settings by US national security agencies.

However, the dispute escalated after the Pentagon sought broader permissions for military uses of AI technology.

Neither Anthropic nor the Pentagon had commented on the latest negotiations on Thursday.

Reporting by Josh White for Sharecast.com.