Why the Left Is Splitting Over AI Doom Politics

American progressives are divided over whether AI should be regulated primarily as a present-day labor, surveillance, and corporate power problem or as an existential threat that could someday end humanity. The split burst into public view in September, when New York City’s Democratic Socialists of America drew backlash for mocking “AI doom” rhetoric just days after a former Anthropic researcher warned about catastrophic risks.

The argument matters because it is reshaping how the left talks about technology, who it aligns with in Washington, and what kind of AI laws could emerge next. It also reflects a broader struggle over whether warnings from frontier AI companies are a useful way to win regulation — or a distraction that helps the industry define the terms of debate.

At the center of the controversy is a growing overlap between two different political conversations. One focuses on worker displacement, surveillance tools, data centers, and the environmental cost of AI infrastructure. The other centers on whether highly capable models could one day escape human control and trigger global disaster. On the American left, those concerns are increasingly colliding rather than neatly separating.

Why is the left arguing about AI doom now?

The left is arguing about AI doom now because AI has moved from a niche policy topic into a live political fight over jobs, infrastructure, power, and safety. The debate intensified as high-profile lawmakers, activists, and former industry insiders began treating the technology as both an immediate source of harm and a potential long-term existential risk.

The spark came after researcher Jacob Coxon, who had recently left Anthropic, publicly warned that the AI industry could help bring about humanity’s end. Two days later, the NYC DSA’s Tech Action Working Group posted a bright pink Instagram graphic arguing that AI hype and AI alarmism were “the same story,” with both serving as distractions from concrete harms happening now.

The post landed hard. It drew more than 25,000 likes, but it also drew sharp criticism from users who argued that sounding the alarm about extreme AI risks is sometimes justified. That reaction captures the core of the split: should the left treat “AI doom” as fringe rhetoric, or as a serious warning that belongs in mainstream policy conversations?

What the New York DSA was trying to say

The NYC DSA’s tech wing is trying to redirect attention from speculative catastrophe toward the real-world damage it says AI is already causing. That includes workplace monitoring, biased automated systems, the expansion of data centers, and the concentration of power in a small number of large corporations.

Alan Deng, a member of the working group, argues that socialist politics should start from the reality of corporate power, not from fears that may sound abstract to ordinary people. In his view, AI is already functioning as a tool of exploitation, and the companies developing it should not be trusted to design the rules that govern it.

“As socialists, we don’t trust these corporations to craft their own regulations,” Deng said, adding that the damage from unregulated and misused AI is already visible across the country.

That framing reflects a longstanding left-wing instinct: when powerful firms ask to be regulated, they are often suspected of trying to shape rules that entrench their dominance. Critics inside this camp say AI existential-risk talk can end up doing exactly that.

How did Bernie Sanders change the debate?

Bernie Sanders changed the debate by bringing AI safety from Silicon Valley circles into national progressive politics. Over the past year, the Vermont senator has spoken about AI not only as a source of job losses and surveillance, but as a technology that could threaten humanity itself.

Sanders also tied AI to another major left-wing organizing issue: data centers. His March bill proposing a moratorium on new data center construction linked the pause to the need for federal regulation designed to protect “the future of humanity.” That language pushed a previously specialized safety conversation into a much broader political arena.

The move is unusual because Sanders has often been associated with labor rights, health care, climate policy, and anti-corporate politics, not with frontier AI governance. Yet on this issue, he has become one of the most visible national figures arguing that the technology needs sweeping guardrails before it advances further.

Why data centers became part of the AI fight

Data centers became part of the AI fight because they are the physical footprint of the industry’s growth. They need enormous amounts of land, electricity, water, and political approval, which has made them a target for environmental activists, local organizers, and community groups worried about utility bills and development pressure.

That issue has helped connect abstract concerns about AI to concrete local campaigns. For many activists, data centers are easier to organize around than distant questions about superintelligent systems. They represent visible infrastructure, local costs, and an industry expansion plan that can be opposed city by city and state by state.

Sanders’s moratorium push gave those activists a national frame. It also drew him into unusual territory: his stance on AI safety echoed arguments often associated with Silicon Valley rationalists and effective altruists, while his criticism of corporate AI overlapped with a broader populist critique of concentrated power.

What do Sanders’s alliances reveal?

Sanders’s alliances reveal that AI safety is no longer a purely partisan or ideological niche. His position has put him in contact — intentionally or otherwise — with figures far outside the progressive ecosystem, including Steve Bannon, the right-wing populist and former adviser to Donald Trump.

That odd overlap underscores a central feature of the AI moment: people across the political spectrum are finding reasons to fear powerful technology companies, even when they disagree on nearly everything else. For some, the concern is monopoly power and labor exploitation. For others, it is catastrophic loss of control. For still others, it is a mix of both.

The political consequences have been immediate. Left-wing candidates around the country have begun referencing AI in increasingly broad terms, often linking existential-risk language to data center opposition, labor concerns, and the need for federal oversight.

On September 9, six candidates associated with the Working Families Party — including four backed by Sanders — issued statements calling for AI regulation to prevent catastrophic outcomes. Their statements suggested that the issue is no longer confined to think tanks or tech podcasts. It is becoming an electoral talking point.

What Alexandria Ocasio-Cortez is signaling

Alexandria Ocasio-Cortez appears to be signaling a more cautious, less explicitly doom-focused approach. She supported Sanders’s data center moratorium and adopted similar language in the House version of the bill, including the reference to humanity’s future. But her public comments have leaned more heavily toward jobs, bubbles, and the physical and economic costs of AI infrastructure.

When asked recently about AI safety, she said more than one thing can be true at the same time. She warned that AI can be dangerous if it is developed without safeguards, while also cautioning that the industry may be inflating a bubble even as it reaches a technological frontier.

That balance matters. Ocasio-Cortez’s rhetoric suggests that left politics on AI may not hinge on accepting or rejecting apocalypse language. Instead, the emerging consensus may be that the technology is simultaneously risky, overhyped, and structurally harmful in its current form.

Why are some progressives pushing back?

Some progressives are pushing back because they believe existential-risk language elevates the wrong problem and empowers the wrong actors. Their argument is not that AI safety is irrelevant, but that the public conversation has been captured by the priorities of the frontier labs and the ideologies that inform them.

One of the loudest critics, researcher Timnit Gebru, has accused Sanders and other left-leaning supporters of AI doom rhetoric of echoing what she sees as dangerous and elitist Silicon Valley subcultures. In her view, the philosophical roots of much AI safety talk — especially effective altruism and rationalism — are not neutral technical traditions but loaded political frameworks.

For critics like Gebru, the risk is that the left ends up amplifying the exact companies it should be restraining. By repeating the idea that AI might destroy humanity, activists may, intentionally or not, lend glamour, urgency, and legitimacy to firms already trying to portray themselves as civilization-defining actors.

Rumman Chowdhury, founder and chief executive of Humane Intelligence, said the industry often benefits when it persuades the public that its models are extraordinary enough to merit extraordinary attention — even when the conversation later lands back on the same practical policy demands critics have been making all along.

Chowdhury, who previously led Twitter’s machine learning ethics, transparency, and accountability work, argued that the spectacle of enormous claims can serve corporate interests by drawing money, attention, and legitimacy toward the companies making them. In her view, the more the industry inflates the stakes, the more investors and regulators are pulled into its gravitational field.

How might AI safety talk help the companies it criticizes?

AI safety talk might help the companies it criticizes because fear can raise their strategic value. If the public believes only a handful of firms are capable of building systems powerful enough to threaten human control, those firms become even more central to the market and to policy negotiations.

That dynamic could matter especially for OpenAI and Anthropic, both of which are widely expected to pursue public-market ambitions in coming years. Greater scrutiny can cut both ways: it can force disclosure and oversight, but it can also help companies argue that they are indispensable participants in a serious regulatory regime.

Chowdhury noted that going public requires extensive disclosure about risk management. From that perspective, an industry that helps define the rules around its own dangers may also be preparing the paperwork and narrative needed for future investor confidence.

That is one reason some left critics worry about the politics of doom. If the companies win the argument that only they understand the dangers, they may also gain more authority to define what counts as acceptable safety, what the public should fear, and which regulations are realistic.

What are the practical concerns the left keeps returning to?

The practical concerns the left keeps returning to are surveillance, workplace displacement, data center expansion, biased automated systems, and the concentration of power in private hands. Those issues are immediate, visible, and tied to everyday life in a way that far-future scenarios are not.

• AI monitoring systems can deepen workplace control.
• Predictive tools can worsen civil liberties concerns.
• Data centers can strain local power and water systems.
• Automation can threaten livelihoods before any “superintelligence” arrives.
• Big AI firms can shape policy by controlling technical expertise and capital.

Pragathi Barusubramanian, a member of the NYC DSA tech working group and a leader in anti-data-center organizing, said that on-the-ground organizing rarely begins with existential fears. Instead, it starts with the sense that AI companies and the firms building infrastructure for them do not share the public’s interests.

Barusubramanian’s view reflects a recurring theme in left organizing: people are more likely to mobilize around concrete harms than speculative ones. The challenge for AI politics is to connect those visible harms to a regulatory strategy broad enough to address them without becoming captive to industry narratives.

What does the Democratic left actually agree on?

The Democratic left appears to agree more than it disagrees on the need for AI regulation. The real dispute is over emphasis, framing, and which harms should drive the agenda. That distinction matters because political movements rarely have unlimited energy, money, or organizational bandwidth.

Alexander McCoy, a Democratic strategist focused on AI, argues that the divide is overstated. He believes the same people concerned about frontier models are also serious about current abuses, and that there is no reason the movement cannot do both.

McCoy said the debate is often framed too narrowly, and that concerns about surveillance, public safety technology, and powerful AI systems should not be treated as competing priorities.

He also pushed back on the idea that critics such as Gebru represent the mainstream of left AI politics. In his view, dismissing frontier-risk concerns as marginal misses the fact that many policymakers and activists now see the issue as a “yes and” proposition rather than an either-or choice.

Phil Aroneanu, executive director of the pro-AI-safety group Irreplaceable, makes a similar case. A former climate activist, he says the left can learn from climate politics, where movements have long tried to address immediate harms from extraction and pollution while also pushing for long-term structural change.

That analogy has power. Climate politics offers a model in which immediate damage and future catastrophe are treated as linked, not competing, concerns. For Aroneanu, AI may require the same layered approach.

What role are candidates and activists playing?

Candidates and activists are turning AI into a campaign issue, a protest issue, and a message discipline issue all at once. The result is a debate that is no longer limited to academics, technologists, or regulators. It is now being fought in local chapters, state capitals, congressional offices, and election-season statements.

That shift is important because it forces political actors to choose language carefully. Saying AI is dangerous is not the same as saying it could destroy humanity. Supporting a data center pause is not the same as endorsing a complete anti-AI platform. But in the current environment, those distinctions are often blurred in public debate.

Inside the DSA, that blur has produced tension. Some members want the organization to focus on the immediate distributional harms of AI and the physical infrastructure supporting it. Others see value in connecting those concerns to broader national and global safety arguments.

Barusubramanian said they are at least glad the issue is finally getting attention. That sentiment may be the clearest sign of the moment’s political reality: the fight is not about whether AI matters, but about what kind of mattering should define the left’s response.

How much public concern is there about AI destroying humanity?

Public concern appears to be real enough that politicians are taking notice. A Politico poll released Wednesday found that two-thirds of Americans think there is at least some risk that AI could destroy humanity. Whether respondents interpret that as a serious forecast or a remote possibility, the number suggests the issue has broken out of niche policy circles.

That kind of polling matters because it changes what elected officials think is politically possible. If voters already accept that AI could pose extreme risks, lawmakers may feel freer to adopt stronger regulatory language — even if they do not fully embrace every claim made by frontier-risk advocates.

At the same time, public fear can be unstable. It may rise after vivid warnings or sensational headlines, then fade when attention shifts back to more immediate concerns like jobs, pricing, or access. That volatility helps explain why both wings of the left are trying to anchor the issue in something more durable.

For one side, that anchor is labor, surveillance, and corporate accountability. For the other, it is existential safety. The political challenge is whether those anchors can support the same coalition.

What happens next in the left’s AI fight?

What happens next in the left’s AI fight will likely depend on whether activists can build a shared regulatory agenda without forcing a single theory of harm. That means deciding which fights are local, which are national, and which are meant to influence the future of the industry itself.

One likely path is a layered strategy: push for restrictions on data centers, stronger labor protections, surveillance limits, transparency mandates, and safety rules for frontier models. Another is a more confrontational approach that treats superintelligence as a category that should be banned outright.

The DSA convention resolution now under consideration points in that more aggressive direction. It would back Sanders and Representative Greg Casar’s effort to ban superintelligence entirely — a policy that would almost certainly alarm major AI labs and intensify the debate over what the left is trying to accomplish.

For now, the largest takeaway may be that the left is no longer arguing about whether AI regulation is necessary. It is arguing about how to frame the danger, which institutions to trust, and whether the future should be imagined through the language of labor struggle, public safety, or civilizational risk.

Key figures in the debate

The left’s AI split is ultimately a fight over political priorities in an era when technology is moving faster than regulation. For some activists, the lesson of AI is that corporations are already harming workers and communities, so the response should be rooted in present-day power. For others, the fact that these systems may someday become vastly more capable means the left cannot afford to ignore the future either.

That tension is unlikely to disappear soon. If anything, as more candidates, organizers, and lawmakers enter the debate, the argument over AI doom may become one of the defining internal disputes of the broader progressive movement.