Bipartisan Majority Wants Mandatory AI Safety Reviews, New Poll Shows

Broad Support for AI Safety Testing

An overwhelming majority of likely voters across party lines want powerful AI systems to undergo mandatory formal safety reviews before public release, according to a new survey from the AI Policy Institute. The findings represent a significant departure from current Trump administration policy, which makes such reviews voluntary for AI companies.

The poll, conducted among 1,007 likely voters on June 11-12, found that Republicans expressed more enthusiasm for government-led safety testing than Democrats, though majorities in both parties supported mandatory reviews. Over 60% of respondents from both parties said the federal government—not AI companies—should set clear safety standards and evaluate compliance.

Why it matters

This polling data arrives as the Trump administration navigates contentious decisions about AI model releases, having recently forced OpenAI to limit distribution of GPT-5.6 and cleared Anthropic's Mythos 5 for restricted access. The bipartisan appetite for stronger oversight suggests political space exists for federal AI regulation, even as administration officials worry such rules could hamper American competitiveness. The shift in Republican attitudes toward government intervention on AI—now exceeding Democratic support—could reshape the legislative landscape.

Regulation Over Prohibition

When presented with policy choices, voters demonstrated nuanced preferences. Two-thirds of respondents said they preferred having AI systems with safety guardrails over banning them entirely. However, when asked to choose between unregulated AI or an outright ban, voters strongly favored prohibition.

Over 80% of both Democratic and Republican participants said AI companies should not build systems smarter than humans until they can demonstrate control over those systems. Almost three-quarters of all respondents predicted AI will become a more important political issue in the future.

Data Centers and Oversight

The survey also addressed the contentious issue of data center proliferation. Forty-seven percent of respondents said they would allow data centers if AI systems being developed had safety requirements and security standards, while 38% favored banning them entirely. America's data center boom currently faces over 300 bans and moratoriums nationwide, with opponents blocking or delaying projects worth nearly $130 billion this year, according to independent researchers.

Shifting Party Dynamics

The poll reveals a striking reversal in partisan attitudes toward AI regulation. In 2024 Pew Research Center polling, 70% of Republican-aligned voters lacked confidence in government's ability to regulate AI, compared with 54% of Democratic-aligned voters. In this month's Pew survey, those dynamics flipped: 74% of Democrats now lack confidence in government AI regulation, compared with 61% of Republicans.

"I think people in the White House who have been trying to push a no-rules-whatsoever perspective are out of step with the American people on the results of the AIPI poll," said Peter Wildeford, director of policy at the AI Policy Network, a policy advocacy group affiliated with the AI Policy Institute.

The survey comes days after federal primary elections that became proxy battles over AI regulation, including New York's 12th Congressional District race, which drew over $40 million largely from political groups representing AI issues.

These details were first reported by NBC News.