Altman Calls Pentagon Deal a Mistake as ChatGPT Deletions Surge 295%

OpenAI CEO Sam Altman has announced amendments to the company's contract with the US Department of Defense, admitting the rushed deal was a mistake. The move follows a massive user exodus from ChatGPT and a dramatic spike in downloads of Anthropic's competing assistant Claude.

«Here is re-post of an internal post: We have been working with the DoW to make some additions in our agreement to make our principles very clear…» — Sam Altman (@sama), original post

Altman acknowledged the company should not have rushed the agreement, conceding that the process appeared «opportunistic and sloppy» despite intentions to reduce tensions and avoid a worse outcome.

Why This Matters

The OpenAI-Pentagon contract, signed on February 28, sits at the intersection of two critically important topics — military AI applications and competition in the AI assistant market. The consumer backlash demonstrated that public sentiment on ethical issues can redistribute market share in the tech industry within days. For the crypto sector, where decentralization and privacy are foundational values, this precedent highlights the reputational vulnerability of centralized AI services.

Timeline and Revised Contract Terms

OpenAI signed the agreement with the Department of Defense on February 28. Just hours before, President Donald Trump ordered federal agencies to stop using Anthropic's products. The directive followed Anthropic CEO Dario Amodei's public prohibition on using his company's AI for mass surveillance and autonomous weapons development.

The revised contract now includes an explicit ban on the «deliberate use of artificial intelligence» for surveillance of US citizens. A separate clause restricts the purchase or use of commercial personal data repositories. The Department of Defense also confirmed that OpenAI's tools will not be made available to intelligence agencies, including the NSA — any such provision would require a separate agreement.

Altman emphasized the importance of protecting civil liberties and stated that AI governance must follow democratic processes. He argued that no private company should decide the fate of the world. He also urged the Pentagon not to add Anthropic to supply chain threat lists and to offer competitors the same terms as OpenAI.

The Scale of the ChatGPT Boycott

User backlash was swift and severe. According to Sensor Tower, ChatGPT app deletions surged 295% on March 1 alone. For context, the average daily uninstall rate over the preceding 30 days was just 9%.

Mobile download trends for Claude and ChatGPT in February. Source: TechCrunch/Appfigures

New ChatGPT downloads also declined — dropping 13% immediately after the contract was signed and another 5% the following day. The app's App Store rating cratered: five-star reviews were cut in half, while negative reviews multiplied 8.5 times on the first night and doubled again on March 2.

«ChatGPT is getting COOKED on the iOS App Store reviews and I think it's deserved.» — Not Greg (@dogecoinmillion), original post

Claude Emerges as the Primary Beneficiary

Anthropic's Claude became the direct beneficiary of the fallout. Downloads of the chatbot rose 37% on February 28 and a further 51% on March 1. As of March 3, Claude held the top position in the US App Store.

Appfigures analysts reported that daily Claude installs in the US surpassed ChatGPT for the first time, with an 88% increase on March 1. The app also topped free iPhone app charts in Belgium, Canada, Germany, Luxembourg, Norway, and Switzerland.

Similarweb estimated that Claude downloads in the US over the past week exceeded January levels by approximately 20 times. However, analysts cautioned against drawing definitive conclusions, noting that other factors may have influenced the statistics.

Industry Response

Beyond the consumer boycott, more than 700 employees at Google and OpenAI signed an open letter calling for a ban on AI use in surveillance systems and for combat operations without direct human oversight. Earlier in mid-February, Bloomberg had reported on SpaceX's plans to develop software for Pentagon autonomous weapons systems.