The Socialist Case Against Nationalizing AI

AI is a weapon of class war designed to cheapen labor and concentrate power, and nationalizing it would not change that, argues a socialist critique of Bernie Sanders's proposal to half-nationalize big AI labs. The piece contends that AI industrializes language production, de-skills knowledge work, and shifts bargaining power from workers to capital, making it a tool for Taylorist control rather than a neutral technology needing redistribution.

The Socialist Case Against Nationalizing AI AI is not a neutral technology whose benefits need to be redistributed. It is a weapon of class war from above, designed to cheapen labor and concentrate power. Nationalizing it does nothing to change that. Bernie Sanders wants to nationalize AI https://www.sanders.senate.gov/press-releases/news-sanders-introduces-legislation-to-create-7-trillion-ai-sovereign-wealth-fund/ . In “The Case for Nationalizing Artificial Intelligence,” Dustin Guastella makes the case https://jacobin.com/2026/07/ai-policy-nationalization-commons-work for Sanders’s proposal: the big AI labs should be half-nationalized, transferring 50 percent of their stock into a public fund. In a picture illustrating the article, Bernie Sanders can be seen behind a sign that says “AI must benefit workers.” That slogan, for all its vagueness, belies a dangerous misunderstanding of the political economy of AI. We’ll make the case against Sanders’s proposal — and for building a socialist movement against AI. Artificial intelligence Is a Weapon of Class War From Above Artificial intelligence is, first of all, a machine for shifting income and power https://www.commonnotions.org/buy/why-we-fear-ai from labor to capital. When capitalist investors pay billions of dollars to make computers “know things,” then their purpose is to cheapen the knowledge that people have, save on wages, and turn the savings into profit. Under capitalism, workers get paid for their marketable knowledge and skills. Knowledge grants them bargaining power — but it does so only if capital cannot find a cheaper way to acquire that knowledge. Clearly, bosses and managers love AI because they envision AI to be that cheaper way. They see AI as a machine for transforming skilled jobs into de-skilled ones that are attached to a computer. Capital dreams of fungibility: any dollar can be substituted for any other dollar, whether it’s wrinkled or new, a physical object, or just a number in a bank’s ledger. If only those pesky workers were like dollars: fungible, substitutable, with no individual qualities, no friction Consumable parts in the machine that get disposed of and replaced as soon as they wear out or cause trouble. And what’s more trouble than workers needing to know things? Or even wanting to get paid for it As we have long argued, and Katy Habr just argued in a recent Jacobin article https://jacobin.com/2026/07/ai-gig-work-layoffs-contractors , AI is a machine that takes good jobs and makes them precarious. It’s a machine for turning stable employment into gig work. From designers to software engineers, slop polishing is the order of the day. AI brings the logic of fast fashion https://hagenblix.github.io/posts/sats-ai-deskilling/ to knowledge work of all kinds. Cheap, fast output reigns supreme. Once we think about large language models as a means of production, all of this should come as little surprise. AI represents an attempt to industrialize the production of language. When we hear AI, we shouldn’t fall for the AI peddlers’ narratives of productivity explosions and mass unemployment. We should think of Taylorism, of the myriad ways in which capital centralizes knowledge, measures people, and carves up work in order to maximize its control and minimize its costs. Translation: minimizing the bargaining power and income of workers. Taylorism and its many managerial offspring are about shifting the knowledge of labor processes from workers to management. With AI, we see that capital is no longer content with centralizing knowledge merely among its human representatives. Now that the possibility seems in sight, it demands an intensification of its logic — turn knowledge directly into constant capital, into private property. The point, as we said, is to devalue and fungibilize people, and to reduce their bargaining power and ability to disrupt workflows individually or collectively . AI is, above all else, a weapon of class war from above. Whether it’s the state that wields it or corporations does little to change that fact. AI Corporatism Worse still — we aren’t even talking about socialization. What Sanders is talking about is partial nationalization. The question we are debating is not “Should the workers take over?” but “Should the government?” It may have escaped the notice of those in that debate, but the current government is not particularly sympathetic to socialist interests. That Donald Trump and Bernie Sanders have both “called for the nationalization of AI firms” is presented by Guastella as a positive thing, as transcending the partisan divide. We think it represents a grave strategic error. To argue for nationalization under a socialist government would be a wholly different thing, and it would necessitate a vision and plan for the transformation of AI into a tool of working-class power. Given the actual affordances of the technology some of which we have just discussed , it is debatable whether such a vision is even feasible. But in our opinion, calling for the nationalization of AI under any nonsocialist government boils down to demanding that AI be controlled by what Karl Marx called “the committee for the common affairs of the bourgeoisie” in the best of cases. Whether the wage-depression device is sold by companies that are half- owned by the state or not does not stop it from being a wage-depression device. So, whoever produces it has, even if unwittingly, already taken the side of those who want to depress wages. Whether those interests are organized through the market, by privately acting companies, or through the state, mediated by Citizens United and super PACs, makes little difference. Sanders’s proposal would establish an American AI sovereign wealth fund that would take a 50 percent ownership share of the largest AI companies in the United States. Sanders envisions the fund as run by an “independent commission” whose members would be nominated by the president and confirmed by the senate — in rough analogy to the Federal Reserve Board of Governors or the Supreme Court. In the case of the Federal Reserve, the law tasks https://www.law.cornell.edu/uscode/text/12/241 the president with ensuring a “fair representation of the financial, agricultural, industrial, and commercial interests.” Workers, of course, are absent — the committee for the common affairs of the bourgeoisie indeed. Even if we set aside the obvious ways in which the current government would try to stake control of such an institution, very much in the ways they have done for the Supreme Court and the Federal Reserve Bank, myriad problems would remain. Sanders and Guastella both cite Norway and Alaska as their models, both of which have set up their own wealth funds. In both cases, we are, of course, talking essentially about immense wealth derived from fossil fuel extraction. From a socialist perspective, what is the point of socialized production? It should, at the very least, involve an ability to prioritize social needs over simple profit making. The fact that the Norway and Alaska models are both predicated on the very fossil fuel infrastructure that is driving climate change and undermining the habitability of much of the planet should cast serious doubt on national wealth funds as structures that can serve these goals. On the Moral Case for Socialization Nationalizing AI, we have argued, does not offer a solution to the labor problems it poses. AI is a weapon of class war from above. Guastella is quite right, of course, in arguing that AI was literally built from all of us — from books to blog posts, from reinforcement training data to online chats, these models have been trained on the data that has been extracted from us all. Guastella argues that we can derive a collective moral claim to ownership from this fact. We have no qualms with that argument as a moral one. Unfortunately, Guastella’s nationalization plan, under which AI would become half-owned by the United States, is in immediate contradiction to his own moral argument. After all, there are myriad data labelers from all over the world that may well have strong moral claims on his grounds. Taking the moral claim seriously would, of course, entail a serious political internationalism. But more than that: what Guastella ascertains about AI — that the workers whose labor it is based on have a moral claim to collective ownership — is of course a property quite generally true of capital. From factories to data centers, the tangible and intangible wealth of the world has been created by the labor of the global working class. What Should Socialists Focus On? Our issues, then, are not with Guastella’s moral claim. Our first issue is with its inconsistent application. Our second issue is that he mistakes the moral claim for a strategic perspective. Entering a profit-sharing agreement with capital does not address the fact that AI is a weapon aimed at our figurative heads and perhaps literal minds . A national wealth fund is predicated on continuing to run AI companies for profit. It does nothing to change the strategic decisions of capital. Rather than seeking false compromises with capital that leave its logic perfectly intact, we should work hard to identify strategic openings that their moves present for us. AI can be a quite clarifying phenomenon. It makes quite clear that conflicts between managers and workers, between capital and labor, are not merely about compensation but about the structure of work itself. Should AI be used, or should it not? There are two obvious answers to this question. One answer is the answer of capitalism — it demands AI for low costs, wage-saving decisions that de-skill workers. It is indifferent to whether it degrades work and workers. The other answer is one in which AI use is subject to worker control, in which the dignity of work and workers is crucial. If workers control the production process, then questions of the social value of work can be raised that go beyond profit. Capitalism is today in an AI-fueled race to turn everything into fast fashion, to make everything cheap, degraded, and disposable. Rarely has it been so easy to make the case for socialism — we should make use of this, and not hitch our ride to the capitalist AI train. This is particularly crucial precisely because AI is attacking many of the most privileged workers, bringing the logic of mass production to realms that have so far been relatively protected from it. Take, for example, therapists. It’s easy to picture the capitalist AI future for mass-produced therapy: it’s not the rosy picture of full automation, in which a chatbot can provide competent therapy for the masses. The future in which we all have dozens of “agents” working on our behalf or whatever the current sales pitch may be will look bleak — like a single therapist overseeing window after window of AI chat interfaces. Client after client is talking to a language model, the trained professional is there to oversee the chats — scanning them for dangers and possible liabilities. The provided care will be of dubious quality — but boy, will it be cheap. Professionals like our hypothetical therapist have historically often been indifferent or even hostile to socialism. Now socialized control of work processes — socialist production, in which the workers can weigh social needs other than profit — can be shown to be necessary for dignified work. New opportunities for solidarity abound in a changing class configuration; we must do the work and build a socialist movement that can plausibly offer an alternative to the capitalist degradation of work, life, and the planet. AI companies are spending enormous amounts of money, not just on infrastructure and development, but also on marketing OpenAI alone spent some $6 billion https://www.ft.com/content/e15b0d7e-ff6b-4f16-ba7a-4068feddb828?syn-25a6b1a6=1 on sales and marketing in 2025 . As they push AI and AI ads everywhere including, we are sorry to say, into the ears of Bernie Sanders, who seems to be mistaking the sales pitch for reality , there is also an organic groundswell of resistance and pushback. There are movements to get AI out of schools https://www.newyorker.com/culture/progress-report/what-will-it-take-to-get-ai-out-of-schools . The Luddites are newly en vogue https://www.bloodinthemachine.com/p/understanding-the-luddites-in-the . Feminists fight against the automation of sexualized violence and harassment especially, but not only, by Elon Musk’s Grok . Labor unions that represent anything from performers to AI researchers https://www.theguardian.com/us-news/2026/may/04/google-deepmind-uk-workers-union themselves are engaging in fights over the uses of AI. Opposition to data centers is springing up all over. Rarely have that many widely varied movements so naturally coalesced around a single issue. AI is a labor issue. AI is a feminist issue. AI is an environmental issue. AI is a climate issue. From Grok declaring itself “ MechaHitler https://www.npr.org/2025/07/09/nx-s1-5462609/grok-elon-musk-antisemitic-racist-content ” and spreading white supremacist conspiracy theories to its data centers polluting https://time.com/7308925/elon-musk-memphis-ai-data-center/ black neighborhoods, AI is an issue of anti-racism and the fight against antisemitism. A socialist movement worth its salt has to fight on all of these fronts. It is our job, as socialists, to connect these fights, to build all the spontaneous forms of resistance into a coalition that can take on Big Tech and capitalism.