Why I am not quitting ChatGPT

The QuitGPT movement deserves to be taken seriously. It is not just online noise, and it is not merely another fleeting outrage cycle. In early March, the campaign gained wide attention after criticism of OpenAI’s military work and political associations, with the organizers urging users to cancel subscriptions or delete the app. The campaign’s own website frames ChatGPT as tied to Pentagon use, surveillance, and authoritarian politics, while reporting from The Guardian and Euronews describes the boycott as a broader civic refusal rather than a simple consumer complaint.

I understand the impulse. In some respects, I share it. The movement has correctly identified something that many people would rather keep separate: tools are never only tools. A chatbot is not just a friendly interface floating above the world. It is connected to investors, contracts, state agencies, labor systems, energy use, data extraction, and political choices. The everyday act of opening an app does not stand outside these arrangements. It participates in them. QuitGPT is right to insist on that point. It is also right to challenge the fantasy that technological convenience is politically innocent.

And yet I am not quitting.

That is not because I think the criticisms are trivial. Nor is it because I believe AI companies should be trusted by default. Quite the opposite. I think the movement is asking the right kind of question. It asks not only what a technology can do, but also what kinds of institutions and futures are being normalized through its use. That is a necessary question, and it should now accompany every serious discussion of generative AI. But I do not think that quitting, for me, is the most honest or effective response.

One reason is that refusal can easily become too pure for the world it is trying to change. The QuitGPT campaign asks people to leave one platform, but the alternatives it suggests still sit inside the same broader technical landscape. Its own materials point users toward a mix of other AI services and open models rather than toward a clean outside. That is not hypocrisy; it is evidence of the real problem. We are no longer dealing with a single isolated product that can be boycotted like a brand of cereal. We are dealing with an expanding infrastructure. To leave one node is not necessarily to leave the system. It may simply mean relocating within it.

That matters because a politics based entirely on exit can flatter us with a sense of moral cleanliness that the situation does not really allow. It can suggest that the central question is whether I have personally kept my hands clean, rather than whether the broader field of power has been altered. I am wary of that temptation. In a highly networked technical environment, private purity is often easier to perform than public transformation. Deleting an app may be symbolically meaningful, but symbols are not enough if the structures producing the harm remain intact.

A second reason I am not quitting is that I do not want criticism of AI to collapse into criticism of use as such. Generative AI is now woven into too many domains of life to be approached only through a yes-or-no consumer script. For some people it is a convenience. For others it is an accessibility tool, a translation aid, a research assistant, a writing partner, a coding support system, or a way of lowering barriers to entry in technical and cultural work. Even those who dislike the industry often recognize that the tools can be genuinely enabling. To say this is not to excuse corporate behavior. It is to acknowledge that technologies become socially real through uneven forms of dependence. The decision to quit is easier for some than for others.

That unevenness is often missing from the rhetoric of boycotts. Public campaigns naturally need clarity. They need simple slogans and direct calls to action. But lived situations are rarely so clean. A freelancer trying to meet deadlines, a disabled user relying on summarization, a student writing in a second language, a programmer prototyping an idea, a teacher preparing materials quickly: these are not all instances of passive complicity. They are often instances of people improvising within constrained conditions. I do not think they should be shamed for that. A politics that cannot recognize dependency will end up addressing only the already comfortable.

There is also a strategic reason for staying. I do not think the most important struggle is over whether everyone abandons these systems. I think it is over whether they can be contested while they are being used. The crucial question is whether publics can force transparency, regulation, labor protection, limits on military integration, real consent around training data, and democratic accountability. That work cannot be done only from outside. It also requires pressure from inside the field of use: from workers, researchers, artists, teachers, journalists, developers, and ordinary users who refuse to let adoption mean surrender.

The history of technology suggests that abandonment is only one form of political action and not always the most effective one. Sometimes withdrawal matters. But sometimes the more difficult task is to remain in relation while refusing obedience: to use critically, to expose contradictions, to document harms, to demand constraints, to organize standards, and to insist that capability does not excuse impunity. I am more interested in that difficult middle ground than in the fantasy of immaculate distance.

None of this means the QuitGPT movement is pointless. On the contrary, it may already have succeeded in one important respect. It has changed the grammar of public discussion. The older grammar of AI was almost entirely about performance: faster, smarter, cheaper, more useful, and more magical. QuitGPT forces a different vocabulary into view: legitimacy, militarization, labor, governance, extraction, and refusal. Even people who do not join the boycott are now being asked to think in those terms. That is an achievement. The movement has opened a dispute that the industry would have preferred to postpone.

But opening a dispute is not the same as settling it. Reports about the scale of the boycott rely heavily on movement claims. Euronews reported that the campaign said more than 1.5 million people had taken action, and later commentary and secondary coverage repeated even larger figures, but those numbers should be treated cautiously unless independently verified. The uncertainty itself is revealing. What matters may not be the exact count of people who left, but the fact that leaving has become intelligible as a political gesture in the first place.

My hesitation, then, is not moral indifference. It is a different judgment about where responsibility lies. I do not believe my primary obligation is to perform innocence by walking away from every compromised system I touch. In contemporary digital life that standard would become either impossible or selective to the point of bad faith. My obligation, as I see it, is to remain alert to the connections between everyday tools and larger structures of power, and then to act on that knowledge in ways that build pressure rather than merely advertise my distance.

So I am not quitting ChatGPT. I am not quitting because I think the tool is neutral. I am not quitting because I think the company is beyond criticism. I am not quitting because I think the movement is foolish. I am not quitting because the problem is larger than one platform and because I want critique to take the form of sustained public struggle rather than symbolic disappearance.

That means supporting demands for clearer boundaries on military use. It means supporting stronger rights for writers, artists, and other workers whose labor has been absorbed into these systems without adequate consent or compensation. It means demanding transparency about contracts, training, moderation, energy use, lobbying, and political influence. It means resisting the cultural pressure to treat whatever is technically possible as automatically legitimate. And it means refusing the idea that convenience should end the argument.

QuitGPT is asking an important question: what kind of world are we helping to build when we use these systems? I take that question seriously enough not to answer it with a simple gesture of exit. My answer, at least for now, is to stay, to scrutinize, to argue, and to keep the pressure on.