Anthropic Sues Pentagon as OpenAI Deal Sparks Mass ChatGPT Exodus

OpenAI revealed additional details about its agreement with the U.S. Department of Defense on March 1, defending the deal against mounting criticism. CEO Sam Altman maintained in an X AMA that the DoD agreed to prohibit domestic mass surveillance and autonomous weapons use, with OpenAI retaining full discretion over its safety stack through cloud deployment and cleared personnel. Altman himself acknowledged the optics were poor, admitting the deal was "definitely rushed."

Critics were swift to challenge those assurances. Techdirtʼs Mike Masnick argued the agreement effectively permits domestic surveillance by deferring to Executive Order 12333, which has historically been used to justify broad intelligence collection. The backlash rapidly translated into user action: the #QuitGPT and #CancelChatGPT hashtags amassed over 36 million views on X, with more than 700,000 users reportedly canceling their ChatGPT subscriptions. Anthropicʼs Claude surged past ChatGPT on Apple App Store charts in what observers described as one of the most dramatic competitive reversals in the consumer AI market.

The episode underscores the reputational risks facing AI companies that navigate government partnerships. While OpenAI frames its approach as technically more robust — arguing its cloud deployment and personnel controls offer greater real-world protection than Anthropicʼs contractual red lines — a significant portion of its user base appears unconvinced, citing the contrast between OpenAIʼs earlier public support for Anthropicʼs stance and its subsequent pivot.

Hundreds of employees from Google and OpenAI signed an open letter expressing support for Anthropicʼs refusal to grant the Pentagon unrestricted access to its Claude AI system. The letter argues that AI companies have a responsibility to maintain ethical red lines even under government pressure — and that Anthropicʼs position on autonomous weapons and mass domestic surveillance represents the correct industry standard.

The move is unusual in tech, where employees rarely take public stances that contradict their own companiesʼ business interests. For OpenAI employees in particular, the letter arrived as their employer was simultaneously finalizing its own agreement with the same Department of Defense. The open letter signals that internal opinion within leading AI labs is far from monolithic on questions of military AI ethics, and that at least some employees feel strongly enough to publicly side with a competitor.

Industry analysts note that the action could pressure both Google and OpenAI to articulate clearer, more binding commitments on what uses of their AI they will and will not permit for government clients. It also adds a human dimension to a debate largely framed in corporate and legal terms — a reminder that the engineers and researchers building these systems have their own ethical stakes in how they are ultimately deployed.