The bipartisan majority of U.S. residents support AI regulation by government, according to recent survey findings, calling into question potential limitations imposed by the federal government’s AI Action Plan.
For years, government has been grappling with questions on how to regulate AI. Under the Trump administration, the strategy has shifted. A proposed provision in the recently passed budget reconciliation bill aimed to limit states’ ability to regulate AI for 10 years, before facing bipartisan opposition and being removed. Now, the administration has released an AI Action Plan that could force states to choose between enforcing state-level regulations they have implemented or losing federal funding.
The new survey, conducted by the University of Maryland’s School of Public Policy’s Program for Public Consultation (PPC), reveals that the majority of both Republicans and Democrats support proposed regulations on AI systems.
“Clearly, Americans are seriously concerned about the current and potential harms from AI,” PPC Director Steven Kull said in a statement. “And while the public is wary of government regulation, they are clearly more wary of the unconstrained development and use of AI.”
The survey found that, while the public acknowledges that regulation could impact the speed of innovation, the majority favor five proposals under consideration by the U.S. Congress to regulate AI. Specifically, this includes three proposed regulations on the use of AI programs for decision-making, two on deepfakes, and the U.S. working to create an international treaty banning autonomous AI-powered weapons.
Regarding decision-making AI, such as that used within hiring or health care, the majority — 84 percent of Republicans and 81 percent of Democrats — favored requiring that AI programs pass a government test before being put to use. The majority also supported allowing government to audit AI programs already in use and requiring companies to fix problems. Eighty-two percent of Republicans and 78 percent of Democrats favored this proposal. AI systems’ effectiveness is based on the quality of their data, and 74 percent favor requiring companies to disclose to the government how decision-making AI models are trained, upon request.
Deepfakes, or AI-manipulated imagery, increasingly pose security risks. This issue was illuminated when the president shared a deepfake video in July. The survey found the majority of respondents — 83 percent of Republicans and 78 percent of Democrats — support prohibiting the use of deepfakes in political advertisements. Eighty percent of respondents support requiring all deepfakes that are publicly shared to be “clearly labeled.”
One respondent stated their belief about this issue, per the announcement: “Purposefully putting out misleading videos and pictures to the public as the truth will be the downfall of any democratic society.”
The use of AI to create or power weapons has been categorized as one of the most extreme risks posed by AI. The U.N. Secretary General has called for nations to ban the development of lethal autonomous weapons, which the U.S. government has opposed thus far; meanwhile, AI is already reshaping military strategy abroad. However, in spite of the U.S. government’s opposition to such a ban, the majority of respondents favored the U.S. establishing an international treaty to ban the development of such weaponry. Seventy-five percent of Republicans and 77 percent of Democrats supported this proposed regulation.
The survey, offered in English and Spanish, was conducted July 30-Aug. 7 with 1,202 adults nationally. It follows a previous PPC survey conducted in March 2024. The full report is available on the PPC website, and those with a vested interest in AI policymaking can complete the survey themselves.