A Pentagon contract simply price OpenAI thousands and thousands of its loyal clients in a single weekend.
ChatGPT, OpenAI’s flagship product, noticed an enormous spike in uninstallations from US customers, with a 295% day-over-day improve on Feb. 28. The surge adopted public backlash after OpenAI signed a army partnership with the US Department of Defense, a deal that rival AI firm Anthropic had beforehand declined over moral considerations.
How a Pentagon deal turned a PR disaster
The cope with OpenAI dates again to July 2025, when Anthropic closed a $200 contract to combine Claude into categorised army networks. In January 2026, Anthropic raised a priority about its use in deadly missions, akin to planning the seize of President Nicholas Maduro.
On Feb. 13, information leaked that the Pentagon was livid with the corporate, accusing it of making an attempt to “veto” army operations primarily based by itself personal ethics. By Feb. 24, Defense Secretary Pete Hegseth issued a deadline to conform to an “All lawful Purpose” by Friday or face penalties.
Two days later, Anthropic CEO Dario Amodei replied with a refusal, which led to President Trump and the Pentagon banning and blacklist Anthropic, designating them a provide chain threat, that means authorities businesses may now not do enterprise with Anthropic.
In a weblog publish, Anthropic defined that the ban was primarily based on their refusal to adjust to two essential calls for, which contain utilizing a customized ChatGPT occasion by way of GenAI.mil for autonomous weapons and mass surveillance of American residents.
Right after this, OpenAI swept in and took the deal, accepting the phrases that Anthropic didn’t.
But only a few hours after it was introduced, the information took the web by storm and was not well-received as customers felt betrayed. Reddit threads had been began, urging current customers to “Cancel and Delete” the app.
Altman speaks, admits he fumbled
OpenAI CEO Sam Altman took to X to acknowledge that the rollout was badly dealt with and to make clear some particulars in regards to the deal. “We shouldn’t have rushed to get this out on Friday”, he wrote.
He additional introduced plans to amend the contract with the federal government, including that the AI methods wouldn’t be used for home surveillance of US residents.
He went on so as to add that the US authorities ought to reverse its determination to maintain out Anthropic and Claude from working with official businesses, describing it as a “very bad decision”.
The numbers look ugly for OpenAI
Market intelligence supplier Sensor Tower information, first reported by TechCrunch, revealed that past the staggering 295% spike in uninstalls, general US downloads for the ChatGPT app fell 13% day over day on Saturday and dropped one other 5% on Sunday.
Users’ sentiment was mirrored within the app opinions. One-star opinions stored pouring in until it surged to a 775% on Saturday, then to a different 100% on Sunday, in line with Sensor Tower. Meanwhile, 5-star opinions fell sharply, over 50% throughout the two days
This is in stark distinction to final 12 months, when the app was probably the most downloaded app on the Apple App Store for 2025.
Claude’s finest weekend since launch
While OpenAI and its CEO took to social platforms like X to do some injury management, its competitor, Anthropic, quietly skilled considered one of its finest weekends in years. Anthropic’s Claude cellular app noticed a 37% day-over-day spike in US downloads on Feb. 27, adopted by one other 51% bounce on Saturday.
At the tip of the weekend, Claude’s app rose greater than 20 ranks to change into No. 1 on the US app retailer, up from No. 2 the earlier week.
What this second truly means for AI
This occasion underscores a incontrovertible fact that the AI trade has lengthy underestimated: customers pay shut consideration to…
