Left‑Wing Split: Present Harms vs. Frontier Risk in AI Policy
NYC DSA Tech Action Working Group: “AI alarmism and AI hype are the same story: both keep your attention on a science-fiction future to distract you from the real harms experienced in the present.”
In early September 2026, a string of high‑profile moments made a split on the American left visible. On September 8, researcher Jacob Coxon warned publicly that advanced AI could pose existential threats. Two days later, on September 10, the New York City Democratic Socialists of America’s Tech Action Working Group posted the quote above on Instagram. The post got more than 25, 000 likes and quickly became shorthand for one side of the debate.
The split is straightforward in its outlines. One faction, community organizers, municipal progressive coalitions and many on the labor‑focused left, insists that policy should focus first on current, tangible harms: surveillance tech (including Flock cameras), automated job displacement, corporate opacity, and the local impacts of energy‑hungry data centers. The other faction, Sen. Bernie Sanders and a set of safety advocates and left‑leaning organizations, argues the left must also push for “frontier governance”: rules targeted at the most advanced model capabilities, development pauses, and limits designed to reduce the long‑term risk of catastrophic outcomes.
What “frontier governance” means
For clarity: “frontier governance” refers to policy aimed at the cutting edge of AI capability, such as licensing or pausing development above specified capability thresholds, mandatory audits for frontier models, and federal rules limiting certain scaling paths. It differs from harm‑focused regulation (labor protections, privacy laws, municipal permitting) because it is defined by technical capability and systemic risk rather than by a specific present harm to a class of people.
The timeline and the players
That early‑September sequence made the schism visible. Sanders introduced a data‑center moratorium bill in March 2026 (S.4214), explicitly tying a pause on new construction to the need for federal regulation to protect “the future of humanity” (see the bill text on Congress.gov). Sanders has also described AI as “the most transformational technology in the history of the world.”
On September 9, six candidates affiliated with the Working Families Party, including four backed by Sanders, released statements calling for Washington to regulate AI to prevent existential threats. Meanwhile, critics inside the technopolitical left pushed back. Researcher Timnit Gebru posted on X, criticizing Sanders and others for aligning with “literal doomsday eugenicist cults” (her post is available at x.com/timnitGebru/status/2095552930707538134).
Organizational dynamics are complicating matters. A resolution before the national DSA convention would endorse Sanders and Rep. Greg Casar’s bill to ban superintelligence and temporarily pause advanced AI development. That vote shows the debate is not merely rhetorical but partisan and institutionally consequential. And new groups have entered the field: Alexander McCoy helped found a left‑leaning pro‑safety organization called Irreplaceable, which McCoy says received initial seed funding from the Center for AI Safety, and Phil Aroneanu serves as its executive director.
Why this split matters politically
The debate shapes the left’s policy priorities and the coalitions it builds. Prioritizing present harms points to municipal bans, data‑use limits, enforceable worker protections, and energy and permitting rules for data centers. Prioritizing frontier governance pulls resources toward licensing regimes, federal capability thresholds, and potential development pauses.
Both camps frame their approach as protective of working people and democracy. They disagree over where political attention and organizing energy should go. Pragathi Barusubramanian of the NYC DSA Tech Action Working Group put it bluntly: “When it comes down to what’s happening on the ground … the conversation about existential risk doesn’t come up at all.”
Supporters of the broader approach counter that addressing long‑term systemic risks needn’t come at the expense of present harms. Alexander McCoy, a Democratic strategist and Irreplaceable co‑founder, said, “The same people concerned about frontier governance and the development of superintelligence are also addressing those current harms … We should address all these other important factors. Those frontier AI abilities are true and increasing in power.” Phil Aroneanu compared it to climate organizing, where activists simultaneously fought immediate local harms and campaigned against long‑term systemic risk.
Why companies are watching (and sometimes steering)
How the left and lawmakers frame AI matters for markets. Executives and investors pay attention to whether regulation will be written around narrow, capability‑based “frontier” definitions or around a menu of present harms. Rumman Chowdhury, CEO of Humane Intelligence, warns that extreme rhetoric can be weaponized, “Part of IPOing is sharing a lot of diligence documentation on risks and how risks have been addressed … This way, they can say, there is a full regulatory regime that they are part of establishing in this industry to guard against the worst risks.”
If federal rules narrowly target specified frontier benchmarks, incumbents with engineering scale and regulatory teams may be better positioned to comply, shape standards, and pass due diligence hurdles ahead of public markets. Critics fear that could ossify advantages for deep‑pocketed labs while drawing activist and regulatory energy away from everyday harms that affect working‑class communities now.
Public opinion and unlikely alliances
Public concern amplifies the political stakes. A Politico poll released the week of September 18, 2026 found that roughly two‑thirds of Americans think there is at least some risk that AI could “destroy humanity.” That widespread concern helps explain why frontier governance moved from niche technical forums into mainstream political debate, and why alliances form across odd lines when existential worry crosses party boundaries.
What leaders should do now
For company executives, civic leaders and organizers the split means preparing for simultaneous pressure from municipal regulators and federal rule‑making that targets more advanced capabilities. A practical, defensible approach is to build a “dual‑pressure” compliance playbook that treats both streams seriously.
Concrete, executable steps:
- Transparency standards: Insist on model documentation (model cards), provenance for training data where legally possible, and third‑party audits tied to procurement contracts or vendor certifications. Use existing frameworks, such as NIST guidance, to shape vendor requirements.
- Worker and community protections: Contract clauses that require automation impact assessments, retraining budgets or transition funds for displaced employees, and enforceable community benefit agreements tied to large data‑center permits.
- Municipal permitting and energy review: Push for energy and environmental review processes for large new data centers, like municipal EIA‑style reviews and public disclosure of sourcing, so local costs and externalities are addressed while federal AI rules are debated.
- Engage federal definitions early: Participate in public comment and standards work on any proposed “frontier” thresholds, and advocate for metrics (compute usage, benchmarking behavior, interpretability requirements) that are technically meaningful and do not unduly favor incumbents.
- 90‑day playbook for C‑suite: Map vendor dependencies. Require vendor model cards and SOC‑type audits within 90 days. Run an automation impact assessment for core business lines. Brief boards on scenario plans for both immediate regulatory changes and possible capability‑targeted rules.
Short takeaways and questions
- Are left‑wing activists united on AI regulation?
No. There are clear fault lines between organizers focused on present harms, surveillance, labor displacement, and data centers, and lawmakers and safety advocates who push to include existential‑risk and frontier governance in the agenda.
- Does focusing on existential risk distract from immediate problems?
Some organizers say yes, arguing that doomsday framing diverts attention and resources from harms people experience now. Others, including some safety advocates and strategists, say the two priorities can and should be pursued together.
- Could frontier labs benefit politically from existential‑risk narratives?
Industry insiders and critics warn it’s possible: emphasizing extreme risks can attract attention and help define regulatory regimes that incumbents are better placed to influence and comply with, especially as firms prepare for IPOs and investor scrutiny.
- How much do voters care about existential AI risk?
A Politico poll released the week of September 18, 2026 found roughly two‑thirds of Americans think there is at least some risk that AI could “destroy humanity, ” so the issue has broad public salience beyond technical circles.
- Will the DSA officially back frontier governance legislation?
As of mid‑September 2026, a resolution at the DSA convention proposing support for Sen. Sanders and Rep. Greg Casar’s bill (which would ban superintelligence and pause advanced AI development) was pending; the organization’s internal debate remains unresolved.
The split on the left over how to talk about and regulate AI is a live political reality. For business leaders and policymakers, the practical move is not to pick one narrative over the other. Design governance and compliance that respond to immediate harms and to possible capability‑driven rules. That protects employees and communities now and keeps organizations ready for a federal regime that may define and police tomorrow’s frontier.