Ethics just showed up as a line item in the app store charts. On Saturday, February 28, 2026, ChatGPT app uninstalls in the U.S. jumped 295% day-over-day as consumers reacted to news of OpenAI's deal with the Department of Defense. Sensor Tower's figure is dramatic against ChatGPT's typical 9% day-over-day uninstall rate over the prior 30 days. In the same window, Anthropic's Claude hit number one on the U.S. App Store after publicly declining to partner with the defense department.
For anyone managing a brand's presence inside AI assistants, the takeaway is not the politics. It is how fast the assistant users choose can change, and how quickly perception moves the market.
Key takeaways
- ChatGPT U.S. uninstalls surged 295% day-over-day on February 28, 2026, versus a 9% baseline, after news of the OpenAI Pentagon deal. - Sentiment cratered in parallel: ChatGPT 1-star reviews jumped 775% on Saturday and another 100% on Sunday, while 5-star reviews fell 50%; U.S. downloads dropped 13% Saturday and 5% more Sunday. - Claude surged as the alternative: U.S. downloads rose 37% Friday and 51% Saturday, it reached number one in six additional countries, and for the first time its daily U.S. downloads passed ChatGPT (Appfigures). Similarweb pegged Claude's weekly U.S. downloads near 20x January. - The GEO implication: users treat AI assistants as substitutable and switch on values, so brand visibility cannot be a single-platform bet on ChatGPT. It has to be tracked across ChatGPT, Gemini, Claude, Perplexity and more. - Brand perception is now a measurable market force. If a company's reputation can move download charts in 48 hours, how your brand is described inside AI answers is a business metric, not a soft one.
What the data shows
The behavioral shift ran deeper than a single uninstall spike. ChatGPT downloads fell 13% on Saturday and another 5% on Sunday, and the reviews told the same story: 1-star reviews up 775% then up another 100%, 5-star reviews cut in half.
Claude absorbed the flow. Downloads climbed 37% on Friday and 51% on Saturday, it took the number one free iPhone spot in Belgium, Canada, Germany, Luxembourg, Norway and Switzerland, and per Appfigures its daily U.S. downloads passed ChatGPT for the first time. Similarweb estimated Claude's weekly U.S. downloads at roughly 20x January levels, while noting the surge may reflect factors beyond the controversy.
Whether this is a temporary protest or a durable shift is the open question. Either way, it is the first quantifiable evidence that users will switch AI assistants over a company's values.
What this means for GEO
Two things follow directly, and both reshape how brands should think about AI visibility.
First, the assistant market is fluid. A single weekend rearranged which app sat at number one. For brands, that kills any strategy anchored on one engine. If your entire GEO effort optimizes for ChatGPT and a chunk of your audience migrates to Claude, Gemini or Perplexity, your visibility does not transfer with them. Being cited well in one assistant says nothing about the others, because their models, training data and citation behavior differ. Presence has to be measured across the full set, and the mix can move faster than a quarterly plan.
