ChatGPT Uninstalls Surged 295% After OpenAI’s Pentagon Deal

On the evening of February 27, 2026, hours after the Trump administration blacklisted Anthropic and designated it a “supply-chain risk to national security,” OpenAI CEO Sam Altman announced a deal to deploy ChatGPT on the Pentagon’s classified networks. The contract Anthropic had just been punished for refusing — OpenAI accepted. By the next evening, the ChatGPT uninstall surge triggered by the Pentagon deal was already rewriting App Store rankings: Claude sat at #1 in the United States for the first time in its history.

The numbers that followed represent the largest consumer revolt against an AI product ever recorded. A 295% spike in ChatGPT uninstalls. A 775% surge in 1-star reviews. And a Claude download wave so large it crashed Anthropic’s servers on Monday.

The ChatGPT Uninstall Surge in Numbers

ChatGPT uninstalls jumped 295% day-over-day on Saturday, February 28, according to Sensor Tower data reported by TechCrunch. ChatGPT’s typical day-over-day uninstall rate over the prior 30 days was 9%. The February 28 spike was more than 30 times that baseline.

The hostility wasn’t passive. One-star App Store reviews surged 775% on Saturday, then grew another 100% on Sunday — while 5-star reviews dropped 50% over the same period. Writing a review requires deliberate effort; this wasn’t accidental churn.

Claude’s US downloads rose 37% on Friday, February 27 — the day Anthropic was blacklisted — and 51% on Saturday. For the first time ever, Claude’s daily US downloads surpassed ChatGPT’s. Fortune reported that Claude hit #1 on the US App Store on Saturday evening, climbing from 42nd place before Anthropic’s Super Bowl ads and 131st on January 30. On Google Play, Claude reached #5 versus ChatGPT at #2 — a smaller gap, but the direction was new.

The surge crashed Claude.ai and its mobile apps for hours on Monday, March 2 — though API services remained unaffected. The backstory behind Anthropic’s standoff with the Pentagon, which began with the February 24 Hegseth-Amodei meeting, had been building for weeks. The consumer response detonated in 48 hours.

The Deal Altman Called ‘Opportunistic and Sloppy’

The timeline from February 24 to February 27 turned a contract dispute into a political crisis. On February 24, Defense Secretary Pete Hegseth gave Anthropic CEO Dario Amodei a deadline: allow unrestricted use of Claude “for all lawful purposes” by 5:01 PM ET on Friday, February 27, or face consequences. Anthropic refused on February 26.

February 27 then unfolded in three acts. That morning, Trump posted on Truth Social directing “EVERY Federal Agency” to “IMMEDIATELY CEASE all use of Anthropic’s technology.” That afternoon, Hegseth designated Anthropic a “supply-chain risk to national security,” a label normally reserved for firms with ties to foreign adversaries, and moved to sever Anthropic’s $200 million Pentagon contract. That evening, Altman announced OpenAI’s deal on X.

On March 3, Altman posted his mea culpa. “We were genuinely trying to de-escalate things and avoid a much worse outcome, but I think it just looked opportunistic and sloppy,” he wrote, as reported by CNBC. He said the deal was “definitely rushed” and he “shouldn’t have rushed” the Friday announcement.

The contradiction at the center of this story: Altman told Fortune that enforcing the supply-chain risk designation on Anthropic “would be very bad for our industry and our country” — then accepted the contract Anthropic was punished for refusing. His own alignment researcher, Leo Gao, publicly called the safety provisions “so obviously just ‘all lawful use’ followed by a bunch of stuff that is not really operative except as window dressing.”

The internal numbers were just as damaging. Ninety-six OpenAI employees signed a letter opposing the deal before it was announced. More than 600 employees across Google and OpenAI signed a cross-company letter titled “We Will Not Be Divided” supporting Anthropic’s stand. Research scientist Aidan McLaughlin posted on X: “I personally don’t think this deal was worth it.” OpenAI and the Pentagon agreed on March 3 to strengthen the contract with explicit surveillance protections — but that language had not been formally signed as of publication.

Dario Amodei’s response carried further in fewer words: “Threats do not change our position: We cannot in good conscience accede to their request.”

Anthropic’s Four-Move Playbook

What makes Anthropic’s position so effective isn’t the Pentagon refusal alone — it’s that every move over the prior three weeks had already built the competitive funnel. On February 8, Anthropic’s Super Bowl LX ad explicitly promised Claude would remain ad-free — and the very next day, February 9, OpenAI began rolling ads into ChatGPT’s free tier. On February 11, Anthropic expanded Claude’s free tier with compaction, file creation, connectors, and skills. On February 27, the Pentagon refusal turned a safety position into a brand position.

Then, on March 2 — the precise moment of maximum user intent to switch — Anthropic made its memory feature free for all users and launched a competitor import tool. The tool works with ChatGPT, Gemini, and Copilot: users paste a specialized prompt into their current chatbot that exports all stored memories, preferences, and project goals in a single code block, then copy that block into Claude’s settings. Memory had been paid-only since October 2025.

Building a cross-platform import system for three competitors is not a weekend project. The tool’s existence implies Anthropic anticipated a switching moment well before the Pentagon crisis gave them one. An Anthropic spokesperson told TechCrunch that daily signups had quadrupled since January, free active users grew over 60%, and paid subscribers more than doubled in 2026 — figures that preceded the February 28 surge.

The Revolt That Won’t Quit

The organized backlash extends beyond app deletions. The QuitGPT campaign claims 1.5 million participants — counting subscription cancellations, app deletions, social media shares, and signups at quitgpt.org. That’s 7.5 times the 200,000 app deletions from the 2017 #DeleteUber movement. QuitGPT organized an in-person protest at OpenAI’s San Francisco headquarters on March 3, branded “No Killer Robots, No AI Surveillance.”

Important calibration: OpenAI claims roughly 900 million weekly active users — the largest AI consumer base ever assembled. Even 1.5 million participants is a fraction of that base. But the #DeleteUber precedent matters — 200,000 deletions contributed to CEO Travis Kalanick’s resignation from Trump’s economic council. ChatGPT is more deeply integrated into professional workflows than ride-hailing, which makes the switching cost higher but the statement louder when users do switch.

The talent threat may outlast the consumer one. Ninety-six internal dissenters plus a research scientist publicly saying the deal “wasn’t worth it” creates a different kind of pressure than App Store reviews. OpenAI can survive angry users. Losing alignment researchers who believe the company’s safety commitments are window dressing is a slower, more dangerous problem.

Anthropic was founded by people who left OpenAI over safety disagreements once before. The pipeline between the two companies only flows one direction.

What the Numbers Can’t Answer Yet

Can Claude convert rage-quitters into retained users? The import tool removes switching friction, but changing AI assistants means retraining habits and workflows. Many ChatGPT users who deleted the app over the weekend may quietly reinstall once the news cycle moves on — and OpenAI’s 900 million weekly active users give it an enormous cushion.

The part that hasn’t gotten enough attention: Altman called Anthropic’s blacklisting “extremely scary” for the industry, then accepted the contract that triggered it. His own alignment team publicly called the safety provisions window dressing. Nearly a hundred of his own employees told him not to sign. That’s not a PR problem — it’s a credibility fracture between OpenAI’s stated values and its revealed preferences. Credibility fractures compound in talent markets long after App Store rankings normalize.

Whether OpenAI’s amended surveillance language gets formally signed in the days ahead, and whether Claude holds the #1 App Store position through the week of March 9, will determine if this is a one-week protest spike or the beginning of habit formation. Anthropic built the funnel. The question is whether anyone stays in it.