{"slug": "dont-just-nationalize-ai-democratize-it", "title": "Don’t Just Nationalize AI. Democratize It.", "summary": "Bernie Sanders proposed nationalizing 50% of major AI firms, but critics argue public ownership alone does not ensure democratic control. The debate highlights the need for meaningful decision-making power for workers and affected communities, not just state or elite control.", "body_md": "# Don’t Just Nationalize AI. Democratize It.\n\nPublic ownership of AI is no guarantee of democracy. We need democratic public ownership to prevent elites from maintaining control of the technology.\n\nWith Bernie Sanders’ recent [announcement](https://www.nytimes.com/2026/06/01/opinion/artificial-intelligence-bernie-sanders.html) of a plan in the works to convert 50 percent of the largest artificial intelligence firms into public ownership, Jacobin has run several pieces on the question of what should be done about the predatory rise and ecologically uncertain future of this new technology. From different analytical angles, [Ben Burgis](https://jacobin.com/2026/07/ai-nationalization-sanders-libertarians-property) and [Dustin Guastella](https://jacobin.com/2026/07/ai-policy-nationalization-commons-work) praise the Sanders proposal and both call for the nationalization of AI firms. By contrast, [Cecilia Rikap](https://jacobin.com/2026/07/ai-big-tech-global-ownership-control) has made the point that ownership does not equal control, and that an American nationalization scheme falls into a “nationalist trap” that would do a further injustice to the rest of the world, whose data also powers these models. She instead calls for “international, democratic control” and new “public institutions.”\n\nI agree with Rikap that what limits many of the perspectives on the Left about nationalization more generally is that “democratization” is so often treated solely as a question of public ownership. The question of who holds the stock of these firms and accumulates wealth from their operations, whether nationally or globally, is crucial. But in this debate so far, the equally difficult question of who actually gets to decide how the technology is used and developed, and by what means those decisions are made, has largely been ignored. Instead, it is assumed that state control and public ownership will generate a system of artificial intelligence that reflects the popular sentiment and will. But this assumption borders on pure fantasy. The Left cannot assume that state control will result in an empowered working class and a deepened democracy. After all, even Donald Trump has [made the case](https://www.ft.com/content/b1ab6106-77e6-4218-9eb4-e44bd56ca400?syn-25a6b1a6=1) for the United States taking a major equity stake in these companies.\n\nDemocratic AI is only possible if the people whose lives are being reshaped by it have meaningful decision-making power over how it is developed and put to use. But who should be the ones to decide how artificial intelligence is developed? Most agree that we shouldn’t leave something so monumental to Anthropic’s Dario Amodei, OpenAI’s Sam Altman, or xAI’s Elon Musk. Neither should we begin with what the engineers writing the code want. Nor should we merely do what will generate the greatest financial return to the artificial intelligence firms’ shareholders. And as much as state control of artificial intelligence might sound appealing, for reasons I outline below, we shouldn’t leave it up to elected officials to govern it. All else being equal, nationalization alone would be insufficient and would only deepen and reproduce the worst harms of AI under the current administration. For the question of democracy, we need to begin with a simple question: Who is being affected by the development of AI? And how deeply?\n\nOf course we all, as a whole, have a stake in its future and should therefore have a say. But two groups in particular bear the costs associated with making a handful of artificial intelligence investors so rich today. The first is workers. AI’s extraordinary valuation is premised almost entirely on the promise of major labor displacement in the future. In what may be the move of a salesman, Amodei of Anthropic has [noted](https://www.axios.com/2025/05/28/ai-jobs-white-collar-unemployment-anthropic) that AI could obliterate half of all white-collar jobs and increase unemployment in the United States up to 10–20 percent in the next one to five years. It is still not clear how these numbers will shake out, even were AI left to develop unimpeded by market forces. But the fact that AI will be used as a laborsaving technology seems unambiguous, even if it does generate some new jobs requiring some new skills. Workers whose jobs are under threat of automation are not mere bystanders; they are quite possibly the largest stakeholders in this technology’s future.\n\nThe second group are those that will suffer AI’s ecological disruptions. Today the massive humming data centers that train and run these models consume enormous amounts of energy and water to do so. In 2025, data centers consumed 448 terawatt-hours of electricity and 4.5 trillion liters of water globally. Furthermore, they generated 189 million tons of carbon dioxide emissions. According to the United Nations University Institute for Water, Environment and Health, annual power consumption is forecast to [double](https://www.reuters.com/business/energy/ai-double-data-centre-power-water-consumption-by-2030-un-researchers-say-2026-06-03/) to 945 terawatt-hours by 2030. The companies behind the AI rollout have kept their cards close to the chest and have disclosed almost no information about the scale at which they are unleashing this new technology onto the planet. This is of course hastening climate change, but also more proximately has resulted in rises in costs of [electricity](https://www.nytimes.com/2026/02/19/climate/data-centers-power-bills.html) for ordinary Americans and a new [movement](https://www.nytimes.com/2026/06/09/climate/data-center-bans.html) to stop the establishment of data centers with over 150 [local moratoria](https://www.interconnectedcapital.com/research/data-center-moratoriums) and several permanent bans passed. Like that first group, this group is overwhelmingly located in the working class.\n\nGiving these two groups, both the workers at threat of being displaced and the broader working-class demos who will share the collective cost of an AI rollout, actual decision-making power over these companies illustrates what political theorists call the *all-affected interests principle*. The notion is quite simple: if an institution or organization makes decisions that deeply affect the interests of some groups in society, those groups should have a meaningful say in those decisions. According to this simple notion, the people being harmed by the development of AI are, by that very harm, the people that should also be able to govern it. It is my core claim that any proposal that does not put meaningful decision-making power into the hands of the working-class demos, whether nationally (because of feasibility) or globally (because of desirability), has not solved the problem.\n\nBut how might we actually get that power into the hands of the people? In Bernie Sanders’s announcement of the American AI Sovereign Wealth Fund Act, he notes that because AI itself was built on the stolen “collective intelligence” of society, including “books, songs, artwork, journalism, computer code, scientific research, videos, conversations, images and ideas spanning generations,” society should own half of it as a public good. If it is the result of the fruits of our own labor, the wealth it generates should benefit us. I won’t delve too deeply into the financial details of the plan, but it involves a one-time 50 percent tax on these companies paid out in stock used to capitalize a sovereign wealth fund that would give the public a major ownership stake in the largest AI companies such as OpenAI, Anthropic, and xAI. In addition to gaining a share of the wealth produced, through the fund’s voting share and equal representation on each of these companies’ boards, the federal government would be given the power “to block decisions that hurt our citizens and to push for policies that help them.”\n\nBut Sanders’s plan has a major limitation with respect to the question of democracy I addressed above with the all-affected principle. Power and control is routed exclusively through federal government appointees, a power that would change hands with each new election. But this way of thinking about public power is largely a residue of Cold War–era debates that strictly contrasted the state and the market as the public versus the private. This distinction is something we have inherited from that long contest between unbridled American capitalism and Soviet central planning. For many on the Left, this is the default framing that structures nearly every conversation about policy alternatives and socialist goals. But the pluralist idea that capitalist democracies are in fact democratic and do not systematically and fundamentally govern in ways that favor elites and capitalist firms is much more Cold War–era saber-rattling about the supremacy of American political institutions than a fact on the ground. In fact, if there is one resounding finding that has come out of the past two decades of research on business power in the United States, it is that firms enjoy incredible disproportionate power in American political institutions, and that this power has deepened profoundly with the weakening of organized labor and the demobilization of working-class movements.\n\nHere is the upshot: any plan to democratize AI, whether nationally or globally, by routing the public’s ownership stake through government officials would be, by default, aligning itself with the billionaires the plan aims to put in check. To put the issue another way, the political desirability of a state-run AI plan is almost entirely dependent on the democratic character of the state itself. While the new excitement that democratic socialists can win elections is heartening, we shouldn’t kid ourselves: the state is already dominated by elites and there are no major countervailing forces outside of it at the moment in either the labor movement or in the streets that can meaningfully pressure it. What representative government has given us is that, in those periods of quiet politics, especially between elections and on issues with low public visibility where movements are demobilized, business influence governs without restraint. Partial nationalization doesn’t circumvent this problem at all, because public ownership is no guarantee of public control. In fact, without the creation of new institutions that actually deepen democracy in the state, public control simply reproduces private control in practice.\n\nThis is not meant to simply say that the state is a tool of capital with no emancipatory potential. It is instead a contradictory terrain of overlapping governing apparatuses that are themselves the institutional results of past struggles. Usually capital is on the winning side, but not always. The National Labor Relations Act of 1935 wasn’t a gift from the Roosevelt administration, it was a law forced by widespread labor disruptions and growing labor organization. Similarly, the civil rights legislation passed in the 1960s that established some basic freedoms for Black workers was not the gift of John F. Kennedy or Lyndon Johnson but the result of the civil rights movement itself. New political institutions that durably empower working class people are never technocratic fixes. They have always been the result of movements that have to win them and defend them. Without that pressure from below no nationalization or democratization scheme could survive the power of business.\n\nIf democratic AI is possible, I don’t believe it will be found in choosing between the state or the market. Instead, it will require movements to build new institutions where those affected are given decision-making power directly. If we know who should have a greater say, *those affected*, then how should they be selected to actively govern? In other words, what should democratic movements actually fight for?\n\nFor reasons I [explored](https://www.versobooks.com/products/755-the-master-s-tools?srsltid=AfmBOopi39BGYMnagxbtEpENO7vohLSabH3u3wn8LmtCBiiMs_R2nCzI) in depth in The Master’s Tools: How Finance Wrecked Democracy (And a Radical Plan to Rebuild It), neither our current system of representative democracy nor the Occupy Wall Street–era trend in direct democracy are really up to the task. Representative democracy, at its root, is an aristocratic procedure for selecting those that govern. As Bernard Manin [notes](https://www.cambridge.org/core/books/principles-of-representative-government/B5F086D557F0A0995D6FEB2730C29EC9) in The Principles of Representative Government, elections “reserve public office for eminent individuals whom their fellow citizens deem superior to others.” And what is worse, this method of selecting who governs is beset by principal-agent problems that have been easily corrupted by both money in politics and the structural power of capital. On the other hand, direct democracy is simply not feasible at scale and suffers from self-selection problems. Think back to the mass meetings at Zuccotti Park, where the people with the time and desire were the ones there rather than a real cross-section of the people affected by Wall Street’s predations.\n\nInstead, we should turn to an old technology to address AI’s democracy problem, the lot (dating back to ancient Athens). With an ownership stake already in hand, a popular assembly filled via a random selection (or sortition) from those affected by the AI rollout, both all of us generally and those most exposed to the harms particularly, would be empowered to deliberate over and make binding decisions about the development of AI. Sortition itself is the mechanism that allows the all-affected principle to be realized in a new political institution, either at a national or global level, that would give working-class people meaningful say over the future of AI. Random selection, statistically, is how you take power out of the hands of billionaires. This would generate much more than a dividend, it would reallocate control.\n\nWhat might such an AI People’s Assembly actually do? Picture a worker at Amazon receiving a notice akin to a jury summons, informing her that she will sit on the assembly for a fixed term to deliberate over binding decisions about how these models are developed and deployed. The assembly meets on one Friday a month, and it is paid work. She is not asked to become an engineer or to master the mathematics of large language models any more than a juror is asked to fully comprehend forensics or the science of DNA evidence. Instead, those with technical expertise in artificial intelligence and its impacts act as consultants to the assembly and are marshaled by neutral facilitators in order to lay out risks, trade-offs, and options that inform the deliberations of the assembly. The participants would learn about an issue, deliberate over a set of options concerning it, and then make decisions about it.\n\nWithin this assembly, standing commissions with veto power could give permanent voice to the constituencies with the deepest stake in AI’s rollout. This might include a worker’s commission to guard against the degradation of labor in the case of displacement and a green commission to hold AI accountable to a development consistent with ecological sustainability at the planetary level. And as democracy is ratcheted up and AI potentially rolled out, the assembly process may be used at multiple scales: decisions about the terms on which a new data center is built should prioritize the voice of the community that is going to host it, decisions about just transitions in entire economic sectors that incorporate AI tools should give greater voice to the workers in those sectors impacted, decisions about AI’s aggregate planetary footprint, whether ecological or with respect to the economic sectors of the world system, would require a body that transcends any given nation-state. But the all-affected principle stands; within these democratic bodies voice is given to those most affected.\n\nBut these assemblies would need legal teeth. This is where an artificial intelligence charter may come into place. In the United States, any AI company operating above some defined size or computing capability might be required to obtain a federal charter to do business, much like banks already do. That very charter could mandate democratic governance mechanisms along the lines I’ve laid out. The model already exists as legislation under consideration in Congress. The Public Banking Act of 2023 introduced by Representatives Rashida Tlaib and Alexandria Ocasio-Cortez would require any public bank holding more than $500 million in assets to be governed in part by a democratic assembly of people randomly selected from the bank’s jurisdiction. Democratic governance for an institution as complicated as a bank is not hypothetical: Costa Rica’s Banco Popular is governed at its highest level by a 290-person workers’ assembly drawn from the country’s key social and economic sectors. The arrangement forces the public bank to incorporate popular demands and democratic oversight. An AI charter could operate similarly, by mandating that for AI firms above a certain threshold such a democratic governance mechanism either be embedded in the firm itself, or that the firms in the industry be subject to an external body that would generate binding regulations on them.\n\nDemocratic AI is possible, but it is not merely a matter of who owns it. It is also about who controls it and how.", "url": "https://wpnews.pro/news/dont-just-nationalize-ai-democratize-it", "canonical_source": "https://jacobin.com/2026/07/ai-democratization-working-class-institutions", "published_at": "2026-07-16 14:48:23+00:00", "updated_at": "2026-07-16 14:56:36.548148+00:00", "lang": "en", "topics": ["ai-policy", "ai-ethics", "artificial-intelligence"], "entities": ["Bernie Sanders", "Jacobin", "Ben Burgis", "Dustin Guastella", "Cecilia Rikap", "Anthropic", "OpenAI", "xAI"], "alternates": {"html": "https://wpnews.pro/news/dont-just-nationalize-ai-democratize-it", "markdown": "https://wpnews.pro/news/dont-just-nationalize-ai-democratize-it.md", "text": "https://wpnews.pro/news/dont-just-nationalize-ai-democratize-it.txt", "jsonld": "https://wpnews.pro/news/dont-just-nationalize-ai-democratize-it.jsonld"}}