Pentagon inks classified AI deals with seven tech giants, Anthropic excluded

The Department of Defense has officially reached agreements with seven major technology firms to integrate their artificial intelligence capabilities into the military's most sensitive classified networks. The move marks a massive escalation in the Pentagon's "AI-first" strategy while pointedly excluding former partner Anthropic following a high-profile clash over ethical guardrails.

Editorial credit: gguy / Shutterstock

The Pentagon confirmed on Friday that SpaceX, OpenAI, Google, Nvidia, Microsoft, Amazon Web Services (AWS), and the venture-backed startup Reflection AI have signed deals to deploy their models within Impact Level 6 (Secret) and Impact Level 7 (Top Secret) environments.

The "any lawful use" standard

The core of the new agreements is a requirement for "any lawful operational use." According to Department of Defense officials, these AI tools will be used for battlefield decision support, intelligence synthesis, and complex logistics.

Important to note, the contracts do not grant the tech companies a "veto" over how their models are deployed. This lack of a kill-switch was the breaking point for Anthropic, which had previously been a primary provider for the military.

Anthropic CEO Dario Amodei reportedly refused to sign any agreement that didn't include explicit guarantees against using its Claude models for mass domestic surveillance or fully autonomous lethal weaponry.

The Anthropic "blacklist" continues

The exclusion of Anthropic is more than a missed contract; it is a full-scale regulatory war. In late February, Defense Secretary Pete Hegseth designated Anthropic a "supply chain risk" - a label typically reserved for foreign adversaries like Huawei.

While a Federal judge recently issued an injunction against some aspects of the designation, the Pentagon's Chief Technology Officer, Emil Michael, confirmed Friday that the company remains barred from new military work.

Breaking "vendor lock"

The Pentagon claims the move to a seven-company roster is designed to avoid "vendor lock," ensuring the military isn't over-dependent on any single AI provider. Consequently, newer entrants like Reflection AI, backed by 1789 Capital, reportedly saw their onboarding process slashed from 18 months to less than 90 days.

The deal was not without controversy. Earlier in the week, over 600 Google employees signed an open letter urging CEO Sundar Pichai to reject the classified deal, citing fears of "irreparable damage" to the company’s reputation.

However, unlike the 2018 withdrawal from Project Maven following similar protests, Google leadership moved forward with the signing. A company spokesperson stated that while Google remains committed to responsible AI, providing API access to the DoD is "a responsible approach to supporting national security."

Looking forward

The Pentagon’s GenAI.mil platform has already seen 1.3 million personnel generate tens of millions of prompts in just five months, signaling that AI is no longer a pilot program but a core component of modern American warfare. "These agreements accelerate the transformation toward establishing the United States military as an AI-first fighting force," the Pentagon said in a statement. "American leadership in AI is indispensable to national security."