Google signed a new contract expanding Pentagon access to its AI systems after Anthropic refused to allow DoD use for domestic mass surveillance and autonomous weapons. The contract shift creates a precedent for how AI vendors can selectively restrict government use cases.
Anthropic's refusal establishes that AI vendors can contractually block specific government applications without losing all federal business. Google's willingness to fill the gap shows how vendor policy differences create compliance arbitrage opportunities.
General counsel at defense contractors should audit current AI vendor agreements for use-case restrictions. Review whether your government work requires AI tools that explicitly permit surveillance or weapons applications.