US Voters Demand AI Regulation, Bipartisan Push for Safety Controls

American voters have reached an unusual cross-party agreement on AI regulation: they want it, they want it comprehensive, and they want it before something goes wrong. A June 2026 report from the Artificial Intelligence Policy Institute, drawing on a survey of 1,007 U.S. likely voters, documents bipartisan support for government AI oversight at levels rarely seen in contemporary U.S. politics. Congress has yet to pass comprehensive federal AI legislation, leaving a widening gap between what the public says it wants and what lawmakers have delivered.

Bipartisan support for AI guardrails

The AIPI report puts support for AI regulatory guardrails at 82% among Democrats and 77% among Republicans. That degree of cross-party alignment on a technology policy question is notable in its own right. A May 2026 survey by the Annenberg Public Policy Center at the University of Pennsylvania found that nearly two-thirds of Americans believe the government has done too little to regulate AI, with that view running at 77% among Democrats, 72% among independents and 53% among Republicans. Concern about AI governance, on this evidence, is not a partisan position.

What the public actually wants

The AIPI report goes beyond general sentiment to document the specific mechanisms the public wants in place. The most widely supported is a guaranteed “off” switch for AI systems, backed by 88% of Democrats and 83% of Republicans. That demand reflects something concrete: a desire for human intervention capacity when a system operates outside its intended parameters.

The figures on controllability track closely. Some 87% of Democrats and 82% of Republicans want Congress to address the risk of AI systems operating beyond human control, and 84% of Democrats and 83% of Republicans say AI systems should not be built until they can be reliably controlled. Mandatory pre-release safety testing drew 72% support from voters in both parties. Taken together, the numbers give policymakers a reasonably precise brief: the public wants safety verification before deployment, not after an incident. For businesses building AI products, that expectation is worth registering alongside the regulatory calendar, pre-deployment testing requirements are also central to the EU AI Act’s high-risk provisions taking effect in August 2026.

The self-regulation problem

The trust picture for AI companies is poor. A June 2026 Politico report, citing polling data commissioned by Anthropicfound that only around 15% of Americans trust AI companies to make decisions about how these technologies should be developed and used. That figure is a significant constraint on the industry’s argument that voluntary commitments and self-imposed guidelines are sufficient. A 2023 YouGov/AIPI poll found that 82% of voters did not trust tech company executives to self-regulate, with 72% preferring slower AI development over faster. The consistency of that finding across multiple surveys over several years makes it harder to dismiss as a momentary public mood.

What this means for business and policy

For companies building or deploying AI, the AIPI report’s findings suggest that proactive investment in safety testing, transparency mechanisms and external audits reflects what the public already expects, not just what regulators may eventually require. The gap between those two pressures is narrowing. Congress has not passed comprehensive federal AI legislation, and the AIPI report’s authors note that the distance between public demand and legislative output remains wide. Whether that gap closes through federal action, state-level measures, or international frameworks like the EU AI Act remains an open question. For more coverage of AI policy and regulation, visit our AI Policy & Regulation section.