This article is written in support of the NYC-DSA City Convention Resolution Stop the AI Arms Race! Members of the NYC-DSA may co-sponsor this resolution via the Convention portal.
“The means of labour passes through different metamorphoses, whose culmination is … an automatic system of machinery … a moving power that moves itself … so that the workers themselves are cast merely as its conscious linkages.”
Karl Marx, 1857
“I have a grandchild who was just one year old and I was thinking, ‘In 20 years he’ll be 21, still just at the beginning of his life. Will he have a life? Will he live in a democracy?’”
Yoshua Bengio, pioneer of artificial neural networks and most cited living scientist, 2026
Anthropic and OpenAI are expanding into New York City.
New York City will lead the struggle to defeat them.
The Democratic Socialists of America ought to take initiative against the AI companies’ project to make human labor obsolete and organize a working class coalition for the democratic steering and ownership of the technology. Our immediate demands should be a pause on frontier AI capabilities research, a moratorium on AI data centers, regulations on frontier AI models, and policies to ensure workers are protected. Our horizon must be the collective control of AI, with its gains shared by all.
It is to raise these demands that members of NYC-DSA have advanced a City Convention resolution: Stop the AI Arms Race!, which at the time of writing is the most co-sponsored resolution.
Rep. Greg Casar and Sen. Bernie Sanders have proposed a bill that would temporarily pause advanced AI development until a federal regulatory regime can be imposed on the AI industry. The bill would also ban the development of AI systems that surpass human capabilities or have the “capacity to overthrow human governments”. Under this legislation, anyone attempting to build such a system, as the leading frontier AI companies are all openly attempting to do, would be subject to 20 years in prison.
Predictably, Anthropic and OpenAI have begun to strenuously lobby against legislation that would imprison their CEOs. Anthropic, which before facing severe regulatory threat was content to push for such guardrails in the abstract, has now started to hem and haw about how recent incidents do not demonstrate real risks.
This legislation comes after several incidents in which AI agents autonomously hacked into other firms without being prompted to do so during internal deployment.
It comes in the midst of a hiring pullback continuing across the entry-level software positions and as AI-driven automation spreads across sectors. It would be too optimistic to assume such automation will be limited to white-collar jobs. Just this year, Amazon’s one-millionth robot was deployed under the direction of DeepFleet–an automated system that coordinates robot swarms to run unmanned fulfillment factories.
While the AI sector is undeniably frothy with fanciful projections, AI-denialismwould have you believe that AI is neither a capable nor dangerous technology, but solely a marketing scheme of vast proportions, that any warnings of AI triggering massive job loss through automation or catastrophic risk through loss of control only serve to pump valuations. Ignore that the industry’s dark money groups deploy sockpuppet accounts to demonize those concerned with catastrophic risk. Ignore that as public pressure increases, Sam Altman now downplays fears of job losses - quite the turn from three years ago when he was saying that, “jobs are definitely going to go away, full stop.” While there is no reason to take Silicon Valley at its word, not rigorously engaging with the dangers of developing AI capabilities is a strategic dead-end.
It is absurd to believe that Bernie Sanders has been somehow tricked by Anthropic, especially when such a thing would imply that a capitalist firm secretly desires the expropriation of its sector and a halt to the research that justifies its valuation.
I, for one, doubt that Sam Altman wants legislatures to pass frontier AI regulations or seize his fortune and power, given that a SuperPAC affiliated with OpenAI has spent millions of dollars against pro-regulation candidates, as well as over a million for Rep. Ritche Torres and Rep. Yvette Clarke. Those denying the threat of AI to dignity and lives of workers thus join the likes of White House AI Czar David Sacks, who called risks from AI the “new climate change” given all of the “Hollywood storytelling” involved.
Casar and Sanders’ bill unfortunately has no chance of passing this session, not while the Democratic establishment controls the Caucus and the Republican Party controls the House. Regardless, this is but one step in the fight against the billionaire class’s attempt to advance more dangerous AI models and achieve the total disempowerment of workers everywhere.
There are many opportunities for the Left, and the DSA in particular, to lead on this most important matter. I am inspired by Melat Kiros, soon to join our ever-expanding bench of federal Socialists-in-Office, who writes in her platform,
“Artificial Intelligence has already begun to revolutionize every aspect of our lives, just like electricity did a century ago. We cannot afford to let the development of this technology continue without the input of everyday Americans, without democratic accountability, and without the cooperation of the global community. AI already touches healthcare, education, jobs, and justice. If we democratize it and ensure its development is transparent, equitable, and publicly accountable, we can unlock its potential to solve massive problems. If we don’t, we risk concentrating power in the hands of a small number of unaccountable corporations and governments in ways that could be catastrophic and irreversible. …
The development of transformative technology cannot be left to the market and the national security state alone. It belongs to all of us.”
The AI industry is a class project to radically concentrate power in the hands of a sliver of elites even at the cost of colossal social disruption, immiseration, and catastrophic harms. They do not care about accelerating carbon emissions from the buildout of data centers, which OpenAI co-founder Ilya Sutskever once said are likely to cover the entire surface of the earth. They actively seek the replacement of human social life with their models, to scrape out of our lives all that is creative and joyous, as even Sam Altman has done in the parenting of his own children.
If we fail to intervene, humanity will be gradually disempowered to machines of their design. The workers left behind, the children who have committed suicide, and the potential millions exposed to novel AI-designed viruses mean nothing in the face of their radical objective: to automate the very task of AI research, and thus to create better and better AIs, of course under their control. Without having to stake a position on whether or not this goal is technically possible, we can recognize that the financial imperatives of these enterprises and the security logic of their imperialist competition with China compel the AI companies to accelerate, advancing a total transformation of social life against all democratic legitimacy and public control.
This must not come to pass! The Democratic Socialists of America has a historical and moral duty to fight back.
Stop the AI Arms Race! commits the New York City DSA to struggle for a moratorium on the construction of data centers, going beyond the existing on-year moratorium to one that creates permanent regulations on data center buildout, something which SIO State Sen. Kirsten Gonzalez commendably fights for. However, this will not be enough. Even if a national moratorium is imposed, AI companies will shift their data center buildout to other countries, especially the United Arab Emirates and other autocracies. Furthermore, increasing algorithmic efficiency in the training of new models has accelerated capabilities growth beyond that which would be expected from compute alone. Data center moratoria will slow the race but will not stop it.
There is also the AI industry of the People’s Republic of China, which is producing open-source models with capabilities roughly 6-months behind that of the United States. If what we care about are the impacts of deployed AI models on public safety, jobs, and mental health rather than the model’s country of origin, then any effort to govern AI must be international.
The only way to prevent more capable AIs from overwhelming piecemeal regulatory structures is an international halt on AI capabilities research. Such a governance structure would necessarily be a federal intervention, and is something that our federal SIOs ought to advocate for in Congress. This would present an existential threat to the AI companies, triggering the collapse of their economic model. Before an international treaty can be negotiated, we should commit ourselves to organizing for municipal, state, and federal labor protections and frontier AI regulations, such as public auditing and whistleblower protections, expanded unemployment insurance, or sectoral bargaining over technology deployment.
At the same time, socialists have a unique opportunity to push for what must be one of the central goals of our movement, as detailed in our platform - “public ownership of the largest corporations and essential industries to ensure democratic control and accountability to the people”. 69% of Americans would concur with seizing 50% of the ownership of the frontier AI companies. There is a wide debate to be had about the shape of socialization, and whether we mean forcing AI research into the public domain or outright nationalizing the entire sector, it is clear that the overwhelming majority of the American people recognize that a technology trained on the common corpus of human art and text, poised to transform the character of work, ought to be held in common rather than by billionaires.
When the people are the masters of artificial intelligence rather than its victims, we will be able to decide to direct it towards useful ends, to the curing of diseases, to the automation of unwanted drudgery. As Rep. Alexandria Ocasio-Cortez said in 2019,
“We should not be haunted by the specter of being automated out of work, We should be excited by that. But the reason we’re not excited by it is because we live in a society where if you don’t have a job, you are left to die. And that is, at its core, our problem.”
With AI under democratic control, we will decide what activities ought to be left to human hands, exalted creative tasks that make life meaningful, and which can be automated to expand flourishing and freedom. DSA members may have quite different views on where those lines lie, with some wishing to ban the technology entirely. Regardless, the point is that those decisions will be ours, not Sam Altman’s, not Dario Amodei’s, not Elon Musk’s. Only with socialization and democratization can AI be deployed towards creating a life of endeavor and discovery, rather than an inhuman, inegalitarian mass-surveillance hellscape.
This is a once in a century political opportunity. We have the chance to utilize one of the great issues of the day to mobilize workers around the democratic control of the economy, to drive a wedge into the Republican Party’s coalition and the Democratic base, and to reorient politics around a question on which our answers are unfathomably popular. We could be at the forefront of the movement, rather than tail-ending a disparate set of professionalized non-profits. It would be more than imprudent for NYC-DSA not to have a clearly articulated position on this issue.
The resolution is but the initiation of a long, chapter-wide conversation on the socialist orientation to AI. Let a thousand flowers bloom. Let us begin the hard work of deliberation without nostrums, informed by scientific evidence, in all channels of our organization. Let us champion structural reforms brought into the realm of possibility by the political-economic crisis of AI, reforms that could decisively shift the balance of power in favor of working people. Let us proclaim that AI will soon no longer be the province of oligarchs. Let us proudly inform our myriad opponents that, in this domain as well, we are on the offensive.
Let us take the initiative, and Stop the AI Arms Race!
And for those interested, a Q&A,
Q: Is the resolution not duplicative with the activities of the Tech Action Working Group?
We commend the Tech Action Working Group’s necessary work organizing against data center construction, empowering tech laborers, and regulating AI deployment.
We believe that this issue is much bigger than one working group, and will require a society-wide response. It would unduly burden the Tech Action Working Group to assume they ought to be responsible for all areas of this response, including labor policy, taxation, environmental protection, and so on. The broad orientation of NYC-DSA towards AI ought to be decided by our most democratic body - Convention, ushering in chapter-wide deliberation.
Of course, we welcome friendly amendments. Should anyone disagree with any component of the resolution’s content, we encourage public discussion, perhaps in this fine journal, so that DSA members can adopt a view in accordance with the science, the evidence, and the strategic imperatives of democratic socialism.
Q: Does recognizing catastrophic risk distract from current harms?
This is a fair concern, but I would raise two notable points of response.
The first is that there is no evidence that worrying about future catastrophic harms or job automation decreases concern with other matters. I strongly believe that all of these threats come from the same bad actors - the billionaire owners of the frontier AI companies, and thus a broad coalition across all those impacted by technology is an absolute necessity.
Second, the harms entailed by the continuation of a process driven by unbridled capitalist competition are not to be ignored. One would not discard the catastrophes imposed by climate change in a decade due to those imposed now. We must not act until it is too late. We must not wait to have guardrails on models until release, but impose public control over the training process itself. We cannot sit by and act defensively until millions are out of work. Socialists ought to organize and agitate now, as to be at the forefront of the political struggle over AI, at the ready, with a policy agenda capable of meeting the moment.
Q: Isn’t AI just a bubble? If we socialize AI,won’t that legitimize it?
While I would personally disagree with the notion that the sector as a whole is a bubble, and would note that the most prominent advocate of this view is quite literally the head of a PR firm for smaller AI companies, it is certainly possible that some of the AI companies will not be able to meet their financial obligations.
The policies we advocate for would not accelerate a bubble. Rather, they would pop it before it could become more systematically dangerous. There is simply no way for AI companies in this view to survive a pause on frontier capabilities research or the uncompensated seizure of 50% of their assets. Nor does expressing concern with AI capabilities increase the valuations of these companies, which lobby incessantly against regulation and sponsor dark money groups to discredit risks of future job loss. What it actually does is increase the risk that these companies are regulated, which would certainly strike some more caution into the hearts of investors. If the NYC-DSA declares that the AI companies will pose catastrophic risks to the public, erode human autonomy, and immiserate and deskill millions of workers, I can guarantee that the venture capitalist class specifically called out by the resolution will not be happy.
Q: What policies could ensure workers benefit from AI?
Socialists understand that the windfall of automation falls to the capitalist class, that workers will either be unemployed and immiserated or told to work ever longer hours. That is, unless we can win policies capable of securing the benefits to labor.
Examples of policies in this area could include the strengthening of unemployment insurance and the creation of a system of wage insurance, the establishment of a social wealth fund, reversing tax neutrality to ensure capital is taxed at an equal rate to labor, Danish-style public investment in retraining, sectoral bargaining and consultation rights over technology deployment, jobs guarantees, the four-day workweek, or expansion of employee stock ownership plans. We recognize a quite broad spectrum of responses, and encourage rigorous debate as to what is viable and desirable. On the matter of socialization, we would be happy for DSA members to explore nationalization, in-kind taxation of AI company stock, forcing research into the public domain, employee co-determination, public financing of AI safety research, or any other approaches comrades and coalition partners may come up with.
Q: Are any of these demands possible on a city or state level?
The most important socialist objectives in the democratic control of AI are federal, and even international, goals. We are confident that the DSA will continue to win power in Congress and, perhaps, the Presidency, and thus strongly believe that federal demands have become necessary.
There is much work to be done at a city and state level as well. We applaud the Tech Action Working Group’s organizing for a moratorium on AI in New York Schools and on new AI data centers. However, there are a great many additional levers that lay unpulled, including, but certainly not limited to, state auditing of frontier AI companies, severance payments for automated jobs, the imposition of more rigorous procurement standards, platform-funded portable accounts, and so on.
Q: Who are the authors of this resolution? What is their factional affiliation?
I am Zachary Jones, a DSA member since the night of Zohran Mamdani’s primary victory, and beforehand a persistent fellow-traveller. I am an MA student at CUNY and have done work for an investigative reporting outlet that tries to keep the AI industry accountable. My co-author is Alexei Gannon, a Computational Neuroscience PhD student who has written extensively on AI, including a useful primer on areas of technical AI safety research. We have collaborated with a wide breadth of DSA members in writing this resolution, across Caucuses and organizing areas and areas of expertise, all of whom care deeply about defeating the plan of the AI companies to disempower the working class.
Neither of us are caucused and this resolution is not a factional project. We welcome everyone, no matter your position on other political questions, to participate in the discourse that will articulate a socialist position on AI.
If you have any further questions on the resolution, or would like to know how you can help further its passage, feel free to contact Alexei or myself.
‘War’, Claudine Bouzonnet-Stella, Jacques Stella, 1657
Thank you from PA DSA comrades. I have been feeling sick at the way this issue has been basically ignored and allowed to fester by so many on the left (not that I've been that active specifically in my chapter recently). It's time to end the arms race and institute an international governance system. Thank you!