Pentagon-OpenAI Deal Hinges on "All Lawful Use" After Anthropic Blacklisting

OpenAI has signed a deal to supply AI models to the Pentagon's classified networks, just hours after the Trump administration blacklisted rival Anthropic for refusing to remove safety restrictions from its military contract. The dispute centers on three words -- "all lawful use" -- a clause the Pentagon demanded that would allow unrestricted deployment of AI models for any legally permitted purpose, without contractual carve-outs for surveillance or autonomous weapons.

The fallout escalated rapidly last week. Anthropic CEO Dario Amodei refused the Pentagon's demands, stating the company could not "in good conscience" agree to unrestricted military use. In response, Defense Secretary Pete Hegseth designated Anthropic a "supply-chain risk to national security" -- a classification previously reserved for foreign adversaries like Huawei -- and President Trump ordered all federal agencies to immediately cease using Anthropic's technology.

OpenAI's contract includes stated prohibitions on domestic mass surveillance and requires human responsibility for use of force. However, critics note these safeguards rely on existing law, which may have significant gaps. Anthropic had argued that "lawful" is not the same as "safe" -- pointing out that bulk government purchase of commercial data for AI analysis would not technically violate current surveillance law.