After being "banned" by Trump, Anthropic and the Pentagon have returned to the negotiating table.

```

Anthropic CEO Dario Amodei is making a final effort regarding the company’s contract with the U.S. Department of Defense, after negotiations between the two sides suddenly broke down last week, putting the AI company at risk of being removed from the military supply chain.

According to the Financial Times on Thursday, Amodei is currently consulting with Deputy Under Secretary of Defense for Research and Engineering Emil Michael, seeking a compromise on the contract terms for the Pentagon’s use of Anthropic AI models. Defense Secretary Pete Hegseth threatened last Friday to designate Anthropic as a supply chain risk enterprise, though this has not yet been formally implemented—the resumption of negotiations is precisely to avoid this risk.

If designated as a supply chain risk, military partners would be forced to cut ties with Anthropic, causing the company to lose the $200 million agreement with the Department of Defense it obtained last July, and closing its access to classified and national security fields. Meanwhile, rival OpenAI last Friday signed an agreement with Hegseth, marking a divergence for the two companies in the U.S. military market.

The restart of negotiations comes against the backdrop of a heated public dispute last week. Emil Michael publicly criticized Amodei, and Amodei subsequently stated in an internal memo that the Pentagon had “lied,” implying Anthropic was being suppressed because it had not sufficiently praised the Trump administration. The leak of the memo made the future of negotiations more complex.

Core of the Disagreement: A Contract Clause About Surveillance

The immediate trigger for last week’s negotiation breakdown was the Pentagon’s insistence on deleting a specific phrase in the contract regarding “bulk data analysis.” According to Amodei’s employee memo, this clause was the only direct reference in the contract to large-scale domestic surveillance scenarios, and prohibiting such use is one of Anthropic’s clearly defined red lines—the other being lethal autonomous weapons.

Previously, the Pentagon pushed for AI companies to allow their technology to be used for any “legal” purpose, which directly intensified the conflict. Amodei wrote in the memo that when negotiations were nearing completion, the Pentagon proposed that if Anthropic deleted this specific phrase, it would accept all other terms—“We find this highly suspicious.” Negotiations then collapsed on Friday, and Emil Michael had publicly called Amodei a “liar” on Thursday and accused him of having a “God complex.”

Leaked Memo Sparks Controversy, Puts Pressure on Negotiations

This internal memo was first reported by The Information, later confirmed by the Financial Times. Amodei used strong language in the memo, declaring that much of the Pentagon’s and OpenAI’s statements were “complete lies or deliberately misleading.” He also implied that Anthropic’s troubles were partly due to the company “not praising Trump like OpenAI CEO Sam Altman.”

Despite strong internal statements, Anthropic opted to return to the negotiating table, reflecting the commercial importance of this contract to the company. The $200 million agreement signed with the Department of Defense last July made it the first company to deploy AI models in classified environments and national security agencies. If supply chain risk designation becomes reality, other military ecosystem partners would be forced to cut ties with Anthropic, with impacts far beyond the single contract itself.

Risk Warning and DisclaimerThe market carries risk; investments should be made cautiously. This article does not constitute personal investment advice and does not take into account individual users’ specific investment goals, financial situations, or needs. Users should consider whether any opinions, views, or conclusions in this article are appropriate to their own circumstances. Investment is at your own risk. ```