the cancel chatgpt movement: what the dow deal means - By Sourav Mishra (@souravvmishra)
why the cancel chatgpt movement blew up after openai's military deal, and what devs are doing now.
the "cancel chatgpt" movement blew up overnight. and it wasn't just about prices or bugs. it was about who gets to use the AI you pay for.
here is what happened, why people care, and what developers are doing now.
the trigger: the dow deal
in late february 2026, two big stories dropped:
-
anthropic said no. the department of war (dow) wanted to use claude for surveillance and weapons. anthropic said absolutely not and walked away.
-
openai said yes. soon after, openai signed a deal with the same department. they claimed the dow would only use it in "lawful ways." but people saw it as training a war machine, and they started canceling subscriptions immediately.
the message was clear: your money was indirectly funding a customer you might totally disagree with.
proof-of-cancellation posts flooded the internet.
why it resonated so hard
this wasn't just a political argument. people were already annoyed for a few reasons:
- bad product updates. many users thought chatgpt was getting worse. it felt wordy and bad at coding. dropping gpt-4o for gpt-5.2 just made people mad. canceling was an easy way to vote with their wallets.
- politics. the earlier "quitgpt" push already put openai under a microscope over political donations and ICE using chatgpt. the dow deal was just the final straw.
- a clear alternative. anthropic drew a hard line against weapons. switching to claude became the easy, "ethical" choice.
so, cancelling chatgpt became both a moral protest and a search for a better product.
what openai said
openai's defense is pretty simple. they give the dow a system, but the dow has to follow strict legal rules.
openai says they still control what they build and manage the risks. they claim it's a bounded government deal, not a blank check.
critics don't buy it. they say once the dow has the tech, it's impossible to control how they use it. plus, openai took the deal that anthropic rejected.
what developers are doing
- jumping ship. a ton of developers are moving their new projects to claude, gemini, or open-source models.
- diversifying. relying on just one api was always risky. now, it's an ethical risk, too. tools like the vercel ai sdk that let you easily swap models are getting super popular.
- going local. some teams are moving to local or self-hosted open-weight models. if you don't pay a subscription, you don't fund anyone else.
- just watching. plenty of people are sticking with chatgpt for now to see how things play out.
the bottom line
the "cancel chatgpt" push worked because it combined a huge controversy (the dow deal) with an existing frustration over product quality.
if you are building AI apps in 2026, you don't have to cancel everything. but you should definitely choose deliberately.
know who you are paying, what they do with the money, and always build your app so you can swap out the AI model if you need to.
key takeaways
- the trigger: openai signed a military deal that anthropic refused.
- the frustration: people were already mad about chatgpt's quality and previous political drama.
- the shift: developers are moving to claude, gemini, or using multi-provider tools like vercel ai sdk so they can switch easily.
- the lesson: don't lock your app into one api. know who you pay and always have a backup plan.
written by sourav mishra, full stack engineer for next.js and ai.
frequently asked questions
q: why did people cancel chatgpt in 2026? openai signed a deal with the department of war right after anthropic publicly refused the same deal over ethics.
q: was it just about ethics? no. people were already frustrated with chatgpt's coding quality and weird updates. the deal just gave them a final reason to leave.
q: what should developers do now? don't lock yourself in. use tools like the vercel ai sdk so you can swap between openai, anthropic, and others easily.
q: did everyone actually cancel? no, a lot of people are staying. but enough people left to make it a massive topic of discussion.