Anthropic–OpenAI Pentagon AI deal triggers QuitGPT revolt

This week’s Pentagon–Anthropic–OpenAI saga is the clearest sign yet that frontier AI labs have become geopolitical assets as much as tech companies. The U.S. Department of Defense effectively forced a choice between unconstrained military use and stronger guardrails; Anthropic walked away and got blacklisted, while OpenAI took the deal and is now fighting a consumer backlash. The fact that a single contract dispute can move millions of users to uninstall ChatGPT and propel Claude to the top of the App Store tells you how politicised the AI stack has become.

For the race to AGI, the episode reframes “alignment” from a purely technical concept to a market and state-power question. OpenAI may gain privileged access to defense data, funding and deployment environments, which could accelerate its capabilities and infrastructure. Anthropic, by contrast, is monetising an ethical brand, converting distrust of government surveillance into user growth and enterprise interest in “safe” AI. That dynamic could encourage other labs to differentiate on values, not just benchmarks.

Long term, the bigger shift is that governments are now willing to reopen signed AI contracts to push for broader powers, as TechCrunch notes, which will make any startup providing AGI-class systems think hard about who they sell to and on what terms. That tension between sovereign demand and public trust is now a core axis of competition.