Google Expands Pentagon AI Access After Anthropic Refuses DoD Contract

WASHINGTON — Google has signed a contract expanding Pentagon access to its artificial intelligence systems after Anthropic refused a Department of Defense request, citing domestic surveillance and autonomous weapons concerns, TechCrunch reported Monday.

The deal underscores diverging approaches between two leading AI companies over the ethical boundaries of military and intelligence applications.

Anthropic, the maker of the Claude AI system, declined the DoD’s overtures on the grounds that the intended uses — which reportedly included domestic mass surveillance and autonomous weapons — fell outside the company’s acceptable use policies, according to the report. The refusal marks a prominent instance of a major AI lab turning down a defense contract on ethical grounds.

Google moved to fill the gap, expanding its existing relationship with the Pentagon through the new agreement. The company has maintained defense and intelligence contracts for years, though its involvement with the military has not been without controversy. In 2018, thousands of Google employees protested Project Maven, a Pentagon program that used AI to analyze drone footage, leading the company to allow that contract to lapse.

The contrasting decisions highlight a fundamental strategic divergence in the AI industry. Google, which has since updated its AI principles to permit certain defense applications, appears to have embraced a broader interpretation of responsible military use. Anthropic, which has positioned itself as a safety-focused AI lab since its founding in 2021, has drawn a firmer line against applications it views as incompatible with its mission.

The development arrives at a moment of intense competition for federal AI contracts. The Defense Department has aggressively expanded its use of artificial intelligence across operations, logistics, intelligence analysis and battlefield systems, creating a lucrative and strategically significant market for AI providers.

For the Pentagon, Anthropic’s refusal and Google’s willingness to step in illustrates both the opportunities and constraints of relying on commercial AI providers whose corporate policies may limit how their technology can be deployed. Defense officials have previously expressed concern about dependency on private-sector AI systems whose terms of service could restrict military applications.

The contract also raises renewed questions about the role of AI in domestic surveillance and the development of autonomous weapons systems — issues that have drawn scrutiny from lawmakers, civil liberties organizations and arms control advocates. Critics of expanded military AI use have warned that autonomous weapons could lower the threshold for the use of force, while proponents argue that AI-enhanced defense capabilities are essential to maintaining U.S. national security.

Neither Google nor Anthropic immediately responded to requests for comment on the specifics of the contract or the terms of Anthropic’s refusal, according to TechCrunch.