AI political cleavages are the fault lines forming around AI policy that traditional left/right partisan alignment does not fully predict. Several orthogonal cleavages cut across conventional party lines, and AI politics has produced cross-coalitions whose members otherwise disagree (for example, Musk, Amodei, and Suleyman on aggressive timelines; Huang and Hassabis on dismissive pushback).
Main cleavages
The principal axes, with representative actors on each side:
| Cleavage | One side | Other side |
|---|---|---|
| Doomer / Normalist | Existential-risk priority (Aschenbrenner, Clark, parts of Anthropic) | AI-as-normal-technology (Narayanan & Kapoor, Hassabis-leaning) |
| Accelerationist / Cautionist | E/acc, full-speed deployment (Sacks, Musk-on-xAI, Andreessen) | Pause / safety-first (Amodei, FLI, AI safety community) |
| Federalist / State-Power | Federal preemption (Executive Order 14365, xAI lawsuit posture) | State innovation laboratories (CA SB 53, Colorado AI Act, state AGs) |
| Big-AI / Open-Weight | Closed-weight frontier (OpenAI, Anthropic primarily, Google) | Open-weight frontier (Reflection AI, Meta Llama, Mistral, DeepSeek, Qwen) |
| Civil-Rights / National-Security | Anti-bias, anti-surveillance, due-process priorities (ACLU, Access Now, state AGs) | National-security uses, Pentagon procurement, export-control hawks (Chief Digital and AI Office (CDAO), DOD, BIS) |
| Capability-Threshold / Procurement-Gated | Pre-release vetting (Anthropic, CAISI, broad-EO framing) | Procurement-driven governance (Procurement-Driven AI Governance, CDAO classified-cohort, narrower-EO framing) |
| Innovation-First / Disclosure-First | Deregulation, no compelled disclosure (Vance Paris speech, Sacks) | Transparency requirements (CA SB 53 transparency, EU AI Act Article 50) |
| Worker-First / Productivity-First | "Protect every worker not every job" (Lawrence Wong, DoL labor policy, China firing-prohibition) | AI productivity gains as primary policy goal (industry CEOs, Action Plan Pillar I) |
Cross-coalitions
The cleavages combine into coalitions whose members align on one axis while differing on others:
- Amodei, Musk, and Suleyman on aggressive timelines, despite the Musk-Altman litigation.
- Hegseth and Sacks on Anthropic-skepticism, for different reasons.
- DOJ, DeSantis, Texas, and the Trump administration on state-preemption, against California, Colorado, Massachusetts, and New York AGs.
- Reflection (open-weight) and DOD CDAO in a procurement decision, an alignment that runs against the usual reading of open-weight positions as anti-establishment.
Bernie-to-Bannon anti-AI populist coalition
A cross-ideological alignment emerged in May 2026 against AI labor displacement and Silicon Valley enrichment, named the "Bernie-to-Bannon coalition" in The AI Backlash Could Get Very Ugly — Lila Shroff (The Atlantic, May 13 2026). Bernie Sanders stated that "AI oligarchs do not want to just replace specific jobs. They want to replace workers." Steve Bannon said Silicon Valley "does not care about the little guy" and ran a podcast titled Stopping the AI Oligarchs From Stealing Humanity. Sen. Josh Hawley (R-MO) questioned whether AI companies' enrichment is "good for children…parents…the American worker." Sen. Mark Warner (D-VA) said he is "enormously concerned that populism from both the left and the right could curb innovation." Blue Rose Research has found populist anti-AI messaging is electorally effective for Democrats heading into the 2026 midterms. The coalition is not formal — there is no joint legislation and no joint endorsements — but the messaging convergence is documented.
Per a Quinnipiac poll cited in The AI Backlash Could Get Very Ugly — Lila Shroff (The Atlantic, May 13 2026), the only US income cohort optimistic about AI in daily life is households earning more than $200K, which the source treats as a structural class basis for the coalition rather than a purely rhetorical one.
Debates and framings
The "pluto-populist" framing
Edward Luce (Why Americans dread AI — Edward Luce (FT, May 12 2026)) coined and popularized "pluto-populist" to describe Trump's alliance with the tech industry: Silicon Valley funding combined with populist political rhetoric directed against safety regulation. Luce's framing also names a dual-positioning critique, arguing that the tech industry is hawkish on China at home (lobbying against domestic regulation as pro-China) while dovish on China abroad (lobbying to sell chips to China). Luce situates this contradiction at the center of the current cleavage map and the Bessent-led US-China AI guardrail talks (Source: New Developments Log/2026-05-14.md).
Correlation among cleavages
The cleavages are not orthogonal in practice; they correlate in complex ways. A federalist-leaning state-power advocate is often also civil-rights-prioritizing; a Big-AI advocate is often also closed-weight-preferring; a procurement-gated frame typically aligns with national-security framings.
Survey evidence on ideological differences
Mass-opinion surveys in 2026 record a left–right difference on AI's distributional effects that is present in several countries and largest in the United States. In the Pew Research Center's 37-country survey, people on the ideological left were more likely than those on the right to expect AI to increase the gap between the rich and the poor in 8 of the 28 countries where ideology was measured: the United States, Spain, Australia, France, Greece, Italy, the United Kingdom and Argentina. U.S. liberals were 31 points more likely than conservatives to hold that view, among the largest ideological gaps the survey measured; Pew notes that its earlier research found people on the left more concerned about inequality in general (Globally, More People Expect AI to Cause Job Loss Than Growth (Pew Research Center, 2026)).
Partisan differences in the United States vary by question. In the AP-NORC/EPIC survey of July 2026, Democrats were more likely than Republicans to expect AI to do more to hurt than help the environment (57% against 40%), the economy (49% against 40%) and society as a whole (57% against 50%), and more concerned about data centers' local effects, while support for requiring data-center developers to pay for the grid upgrades they need was similar in both parties (79% and 76%) (AP-NORC/EPIC 2026 Energy Survey: AI, Data Centers, Climate and Energy).
Positioning ahead of the 2026 midterms and 2028
Both parties' most prominent figures staked out positions in mid-September 2026, and they diverge on whether AI risk is a subject for public policy at all.
Barack Obama told a private fundraising event on 10 September 2026 that AI "is something that is moving very fast in private hands, and if we don't get on top of it, I think can be dangerous," according to a partial transcript released by his office and reported by the New York Times on 13 September. He urged House Minority Leader Hakeem Jeffries to convene a public policy conversation should Democrats win the House on 3 November — "Once you are speaker, I would strongly urge that the Democrats put together a framework for a very public conversation" — and called for candidates in the 2028 presidential race to make AI a central agenda item with a clear plan addressing safety and economic concerns. Obama's office did not respond to a request for comment (Source: reuters.com).
President Donald Trump, asked by reporters in Ireland on 13 September whether the AI industry should slow down or face more regulation, said: "We're leading China in AI. We're the most sophisticated country in the world, and frankly I want to keep it that way because whoever wins AI wins," adding that "you have a lot of very negative forces that are bringing it up that shouldn't be bringing it up and they're bringing up things that won't happen" (Source: reuters.com).
The exchange does not map onto the doomer/normalist cleavage cleanly: neither figure takes a position on capability timelines, and both frame the question through the competition with China — one as a reason to build a framework, the other as a reason not to. It cuts against the Bernie-to-Bannon pattern above in a different way, since the populist coalition's critique is of AI-sector concentration rather than of risk from the technology.
Relation to policy
The civil-rights side of these cleavages is collected on AI and Democracy and AI and Civil Liberties; the national-security side on AI and National Security and Procurement-Driven AI Governance.
Relationships
- related: AI Governance (umbrella) (umbrella).
- related: AI and Democracy, AI and Civil Liberties, AI and National Security, Procurement-Driven AI Governance.
- related: Techno-Federalism, AI Industry Lobbying — The 2025-2026 Political Offensive (super-PAC infrastructure).
- supports: Globally, More People Expect AI to Cause Job Loss Than Growth (Pew Research Center, 2026) — left–right differences on AI and inequality in 8 of 28 countries.
- supports: AP-NORC/EPIC 2026 Energy Survey: AI, Data Centers, Climate and Energy — U.S. partisan differences on AI's effects alongside bipartisan support for developer-pays grid requirements.
- related: Public Opinion on AI — the survey evidence behind mass-level cleavages.