QuitGPT: 2.5 Million People Are Leaving ChatGPT After the OpenAI Pentagon Deal

On February 28, 2026, OpenAI signed a contract with the United States Department of Defense. By the end of that week, a boycott movement called QuitGPT had documented over 2.5 million people pledging to cancel their ChatGPT subscriptions. ChatGPT daily uninstalls spiked 295 percent above their average. Claude overtook ChatGPT in the Apple US App Store rankings for the first time.

This is the largest organised consumer revolt against an AI company in history. Here is a precise account of what happened, what the numbers actually mean, and what it signals for the AI industry.

What Triggered the Boycott

On February 27, 2026, the US Department of Defense publicly disclosed that it had approached both OpenAI and Anthropic about providing AI services for military applications.

Anthropic CEO Dario Amodei responded the same day: "I cannot in good conscience accede to the Pentagon's request." Anthropic declined.

The following morning, February 28, OpenAI announced it had signed the contract. The company published a short statement framing the deal as supporting national security and responsible AI deployment in defence applications.

The contrast was immediate and visible. Within hours, quitgpt.org launched with the tagline "ChatGPT takes Trump's killer robot deal." The site directed users to cancel their subscriptions and listed Claude, Gemini, and Perplexity as alternatives.

The Numbers — What They Actually Mean

The 2.5 million figure is a composite metric. It includes people who cancelled paid subscriptions, free account deletions, and social media pledges. The numbers require some interpretation:

Confirmed paid cancellations: Approximately 1.5 million active paid subscribers cancelled in the first week following February 28. At an average subscription price of roughly $20 per month, this represents an estimated $30 million in lost monthly recurring revenue — annualised to over $360 million.

Uninstall spike: Mobile analytics firms recorded a 295 percent spike in ChatGPT daily uninstalls on February 28 — the single largest recorded uninstall event for any major AI application.

App Store ranking: Claude surpassed ChatGPT in the Apple US App Store free app rankings for the first time during the week of March 1. Claude downloads rose 37 percent on Friday and 51 percent on Saturday of that week.

Executive departure: At least one senior OpenAI executive resigned in the days following the announcement, citing the Pentagon deal as the reason.

Why This Audience Specifically

The people most likely to pay $20 per month for ChatGPT are software developers, researchers, writers, and knowledge workers. This cohort skews toward concerns about AI safety, military ethics, and corporate accountability. It is also the cohort most likely to have heard of Anthropic, to know that Claude exists, and to be capable of switching without significant friction.

OpenAI was not losing casual users. It was losing the users who evangelised the product, who wrote the blog posts about how to use it, who recommended it to their teams and clients.

What OpenAI Said

OpenAI pushed back on the framing. A company spokesperson said the Pentagon contract covers non-lethal applications — logistics optimisation, administrative automation, cybersecurity tooling — and does not involve weapons development or targeting systems.

The company also noted that it had updated its usage policies in 2024 to allow defence applications after previously prohibiting them. Critics argued the policy change itself was the problem — that OpenAI quietly removed a guardrail without public announcement.

The Claude Effect

Anthropic was the direct beneficiary. Claude's App Store surge, its first-ever ranking above ChatGPT in the US, was not driven by a product update or marketing campaign. It was driven entirely by users actively looking for an alternative that had publicly refused the Pentagon contract.

This created a rare situation in consumer tech: a company gained market share by declining business. Anthropic's public refusal became a product differentiator in a way that no advertising could replicate.

What It Means Long-Term

The QuitGPT movement will not permanently shrink OpenAI. The company has over 300 million weekly active users. Losing 1.5 million paid subscribers is financially significant but not existential. OpenAI's $840 billion valuation reflects expectations that dwarf its current subscription revenue.

What the boycott revealed is something structurally important: AI model choice is now a values-based decision for a meaningful segment of users. When AI tools become infrastructure — embedded in daily workflows, creative processes, and professional output — the ethical posture of the company providing that infrastructure starts to matter in the same way that the ethics of a bank or a news outlet matters to its customers.

For developers evaluating which API to build on, which model to call, which company to depend on for the next five years: the QuitGPT movement is a signal that your users may have opinions about that choice.