Last updated: March 4, 2026
What Happened: The Pentagon Deal That Broke the Dam
On February 28, 2026, Sam Altman announced that OpenAI would deploy its AI models inside the Pentagon's classified network. The deal gave the U.S. Department of Defense access to OpenAI's most powerful models for "any lawful purpose," a scope so broad it alarmed even people who had been defending the company for years.
What made the announcement explosive wasn't just the deal itself. It was the timing. Just hours earlier, on February 27, Anthropic CEO Dario Amodei had publicly declined the same Pentagon request, stating he "cannot in good conscience accede" to giving unrestricted AI access to the military. Anthropic walked away from what was reportedly a $200 million contract.
The contrast was devastating for OpenAI's public image. One company said no. The other said yes. And within 48 hours, the internet had a name for what happened next: QuitGPT.
The QuitGPT campaign reports that more than 1.5 million people have taken action through subscription cancellations, social media boycott messages, or signing up via quitgpt.org. The campaign accuses OpenAI of accepting a "corrupt deal" that allows Pentagon use for "killer robots and mass surveillance."
Timeline: How the Boycott Unfolded
Anthropic CEO Dario Amodei publicly rejects Pentagon's demand for unrestricted AI access, walking away from a reported $200 million contract. Amodei warns that "AI can undermine, rather than defend, democratic values."
Hours after Anthropic's refusal, Sam Altman announces OpenAI will deploy models in the Pentagon's classified network. The phrase "any lawful purpose" becomes the focal point of criticism.
QuitGPT launches. The #CancelChatGPT hashtag begins trending across social media platforms. Early cancellation numbers climb past 500,000 within the first 24 hours.
ChatGPT mobile app uninstallations nearly triple compared to the previous Saturday. Anthropic's Claude surges 37% on Friday and 51% on Saturday. Claude becomes the #1 free app on Apple's App Store in the United States.
QuitGPT organizes an in-person protest at OpenAI's San Francisco headquarters. The boycott count passes 1.5 million. Sam Altman tells reporters he "shouldn't have rushed" the Pentagon announcement.
The Pentagon Deal Was the Trigger, Not the Cause
The speed and scale of the boycott surprised even its organizers. But this didn't come out of nowhere. The Pentagon deal was the final straw for a user base that had been growing increasingly frustrated with OpenAI throughout 2025 and into 2026.
Users had already been dealing with months of declining model quality. GPT-5.2, released in early 2026, drew immediate backlash. Sam Altman himself admitted OpenAI "screwed up" the writing quality. Users described responses as bland, overly cautious, and less capable than previous versions.
There was also the ICE controversy. Reports from FedScoop and the Washington Post revealed that U.S. Immigration and Customs Enforcement had incorporated GPT-4 into recruiting and vetting procedures. For many users, the Pentagon deal was confirmation of a pattern they had already been watching.
The subscription value proposition had been eroding for months. Users paying $20/month were getting a product that felt worse with each update, more restricted with each policy change, and now aligned with military applications they fundamentally opposed. The Pentagon deal didn't create the anger. It gave the anger a name and a clear action: cancel.
Where 1.5 Million Users Went After Cancelling
The biggest beneficiary of the ChatGPT exodus has been Anthropic's Claude. The numbers tell the story: a 37% usage surge on Friday, a 51% surge on Saturday, and a climb to the #1 free app position on the iOS App Store in the United States.
The irony is hard to miss. Anthropic's principled rejection of the Pentagon deal didn't just earn press coverage. It earned paying customers. Many former ChatGPT subscribers explicitly cited Anthropic's stance as the reason they switched.
Google's Gemini has been the second major destination, particularly for users who value integration with existing Google services. The QuitGPT campaign itself recommends several ChatGPT alternatives, including Confer, Alpine, Lumo, Google Gemini, and Anthropic's Claude, while explicitly excluding Grok from its recommendation list.
Open-source alternatives have also seen increased interest. Users who were already uncomfortable with the closed-model approach now have an additional reason to explore self-hosted options that don't come with the ethical baggage of corporate AI partnerships.
The Petition Movement: Users Had Been Organizing for Months
The QuitGPT boycott didn't emerge from a vacuum. There had been organized user dissatisfaction with OpenAI for months before the Pentagon announcement. The most visible example is the GPT-4o petition, which gathered thousands of signatures from users demanding that OpenAI restore the original GPT-4o model after the GPT-5 rollout degraded their experience.
These petitions documented a pattern: users who paid for a certain quality of service watching that service deteriorate while the price stayed the same. The Pentagon deal transformed individual frustration into collective action, but the organizational infrastructure, the shared vocabulary of complaint, the subreddits and Discord servers full of disappointed users, all of that already existed.
The user stories collected on this site over the past year paint a consistent picture. Developers getting "ethics lectures" instead of code help. Writers watching their creative tool become progressively more timid. Researchers hitting arbitrary content restrictions on legitimate academic work. Each story is one person's experience, but together they explain why 1.5 million people were ready to leave the moment they were given a reason.
What the Exodus Means for OpenAI and the AI Industry
At $20 per month per subscriber, 1.5 million cancellations represents $30 million in lost monthly revenue, or $360 million annualized. That's a significant hit for a company that was already burning cash at an unsustainable rate.
But the financial damage may be less important than the reputational damage. OpenAI built its brand on the promise of "safe and beneficial" artificial general intelligence. The Pentagon deal, combined with the ICE integration and the ongoing quality decline, has made that promise ring hollow for a large and vocal segment of its user base.
The broader implication is that AI companies can no longer assume user loyalty. The switching costs between AI assistants are low. The products are increasingly comparable. When the only thing keeping users on a platform is habit, a single controversy can trigger mass migration.
For the rest of the industry, the lesson is clear: users care about more than capability. They care about what their subscription dollars are funding. Anthropic's decision to walk away from $200 million didn't just earn moral authority. It earned market share. That may turn out to be the most consequential business decision in AI this year.
The full timeline of ChatGPT's decline from its peak in 2023 to the March 2026 exodus documents every step of how we got here. The legal actions against OpenAI continue to mount. And the user revolt shows no signs of slowing down.