Menu

Categories

Tags

Pentagon signs 8 AI vendors, booting Anthropic from classified networks

May 2, 2026 | Source: nytimes | Google, OpenAI | 152 views 0 comments

The Pentagon has struck deals with 8 tech companies to deploy their AI tools and infrastructure on classified military networks — specifically the secret-level IL6 and top-secret IL7 tiers. New signatories include Nvidia, Microsoft, AWS, and startup Reflection AI, joining previously signed xAI, OpenAI, and Google, plus Oracle, which was added the same day.

The core term: all companies agreed that the Pentagon can use their tech for "any lawful purpose." Anthropic is not on the list. Previously, Claude was the only large model used by the Pentagon in classified environments, but CEO Dario Amodei refused to accept that standard, demanding written guarantees against mass domestic surveillance and lethal autonomous weapons. In late February, the Pentagon labeled Anthropic a supply chain risk, then Trump banned the company from all government contracts. Both sides are now fighting in federal court. Defense Secretary Pete Hegseth called Amodei an "ideological madman" during a congressional hearing.

However, The New York Times reports that at least some of the new signatories, while publicly accepting the "any lawful purpose" standard, privately secured safeguards similar to what Anthropic had demanded, covering restrictions on autonomous drones and domestic surveillance.

Pentagon CTO Emil Michael said open-source models represent a new direction. Nvidia and Reflection AI offer open-source models; Microsoft and AWS provide infrastructure; xAI, OpenAI, and Google offer proprietary models. Michael noted that Chinese open-source models have already infiltrated many companies, so the US needs alternatives.

Despite the Pentagon's rush to replace Anthropic with OpenAI and Google, the transition has hit technical issues. Intelligence analysts are still using old Anthropic models for daily work. Trump ordered a ban on Anthropic by mid-year, but its Mythos model is particularly strong in cyber offense and defense. The New York Times says the ban is almost certain to be withdrawn. The White House is reportedly drafting an executive order to bring Anthropic back into the fold.

Leave a Reply

Your email address will not be published. Required fields are marked *