Proceed with Caution: What Voters Want from Technology Regulation

When voters think about technology, they’re asking a simple question: What’s in it for me? They’re pessimistic about AI, concerned about social media and data privacy, and deeply distrustful of data centers. But beneath that broad unease, voters are most worried about the immediate risks technology poses to them and their families—and they want policymakers to put in place guardrails to protect them.

In July, Third Way and GBAO surveyed 1,000 registered voters nationwide to gauge their feelings about AI, social media, and data centers—and how they want policymakers to respond. We found that voters are far more concerned about the immediate, practical risks of technology than abstract threats or the race to innovate. They want policymakers to protect them from tangible harms, and they’re willing to sacrifice some speed and innovation for greater safety. But they’re not ready to freeze innovation entirely. 

That creates a clear lane for policymakers: take voters’ concerns seriously, put meaningful guardrails in place to protect people, and preserve the benefits of continued innovation. 

1. Voters Are Most Concerned About the Practical, Immediate Impacts of Technology

Voters’ concerns about technology are close to home. They worry most about threats to their families, finances, and personal information—and less about abstract risks like the pace of AI development or competition with China. And when asked what policymakers should do, they prioritize protections against those same tangible harms.

• Privacy concerns are nearly universal. Ninety percent of voters are concerned about how technology companies collect, store, and use their personal data. Concerns about personal data falling into the wrong hands or being used to set individual prices for daily essentials ranked higher than concerns about terrorism, climate change, and data centers using too much electricity.
• Scams, deepfakes, and online predators are top of mind. Two-thirds of voters (67%) think it is very likely that deepfake videos will be used to confuse, embarrass, or blackmail people. Voters are particularly about seniors becoming victims of increasingly sophisticated online scams, which ranked just behind the cost of living rising too quickly. The single highest-ranked concern is children and teenagers becoming victims of online predators. 

• More abstract threats rank much lower. The pace of AI development and the threat of China overtaking the United States as the world’s dominant economic power are much lower concerns. Both issues landed in the bottom half of voters’ concerns, with competition with China coming in dead last.
• Voters want policymakers to protect them from tangible harms. Their top policy priority is protecting victims of scams, deepfakes, and other harmful AI-generated content (74% strong support/89% total support). This was followed by creating safety standards for chatbots used by children (70% strong support/83% total support) and creating safety standards for the most powerful AI models (58% strong support/84% total support).

2. Familiarity with AI Reduces Fear, but Doesn’t Reduce Demand for Regulation

The more voters use AI, the more optimistic they are about it. Daily users see clear benefits from the technology, while occasional users and non-users are much more pessimistic. But familiarity doesn’t eliminate the desire for guardrails: even daily AI users overwhelmingly believe some regulation is necessary.

• Voters overall are skeptical of AI, but daily users are a striking exception. Just 31% of voters have a favorable view of AI, compared to 51% who view it unfavorably. Among daily users, those numbers flip: 55% view AI favorably, and just 30% view it unfavorably.
• The same divide shows up when voters consider AI’s impact on society. Nearly three-quarters of non-users (74%) think AI will have a negative impact on society, compared to 51% of occasional users. Daily users see things very differently: 57% expect AI to have a positive impact on society.
• Sentiments about personal impact are less extreme. While 67% of non-users think AI will have a negative impact on them personally, only 44% of occasional users agree, and 21% are undecided about its impact. Two-thirds of daily users think it will have a positive impact on them personally. 
• But optimism about AI doesn’t translate into opposition to regulation. Just 3% of daily users say no regulation is needed—even fewer than occasional users (5%) and non-users (6%).
• Voters largely agree on what policymakers should prioritize. Across all three groups, voters prioritize protecting kids, creating national standards, and balancing guardrails with innovation.

3. Voters Want Government to Help Protect Kids Online, but Not to Replace Parents

Protecting kids from the risks of technology is a top priority for voters, and there is broad support for government action to keep them safe. But voters don’t want to hand over all oversight to the government. Our findings suggest parents want help protecting their kids, but don’t want to be completely hands-off. 

• The risks technology poses to kids are top of mind. Children and teenagers becoming victims of online predators ranked as voters’ single biggest concern—even above the rising cost of living. And 83% agreed that we need to carefully monitor how AI is used around kids and help children develop a healthy relationship with the technology.
• Voters want parents to remain on the front lines. A majority of voters (52%) think parents should be primarily responsible for keeping their children safe online, compared to 37% who think the government should set basic child safety standards. 
• But some groups want a bigger role for government. A majority of Democrats (52%) think the government should set basic safety standards for tech companies to address child safety. Dads strongly favor parental responsibility (60%), while moms were more divided: 44% said parents should be primarily responsible, while 48% said government should step in. 

• There is broad support for concrete government protections. Eighty-three percent of voters support creating safety standards for chatbots used by children—the third most popular policy proposal overall.
• One thing all voters agree on: tech companies can’t be trusted to police themselves. Only 7% of voters thought tech companies can be trusted to set their own child safety policies.

4. Voters Dislike Data Centers, but Other Technology Concerns Take Priority

The message from voters is clear: data centers are deeply unpopular. They worry about their impact on water and electricity and aren’t interested in making it easier to build them. But when forced to weigh data centers against other technology concerns, voters prioritize more immediate threats like scams, data misuse, and online predators.

• Data centers are deeply unpopular—even more than AI. Just 18% of voters have a favorable view of data centers, compared to 31% who view AI favorably. Negative opinions cut across age, party affiliation, and community type. 
• Even AI’s biggest fans don’t like data centers. Half of daily users have an unfavorable view of data centers, despite being much more optimistic about AI itself.
• Voters are willing to make data centers harder to build. Sixty-two percent of voters say we should make it harder to build data centers because they strain supplies and raise costs. And 65% would support pausing data center construction until AI regulations are developed.
• But data centers aren’t a top priority. Concerns about data centers ranked in the bottom half of the issues we tested, behind concerns about data misuse, frauds and scams, and online predators. 
• The same pattern holds when voters consider policy responses. While a pause on data center construction did get majority support, the proposal ranks near the bottom of the policies tested. Voters prioritize training small businesses to use AI, protecting artists from AI models using their work without consent, and creating strong safety guardrails.

5. Trust in Institutions is Low, but Voters Do Want Regulation

Voters have little faith in the institutions that could oversee technology—from government and political parties to tech companies themselves. But that distrust hasn’t dampened demand for action. Voters overwhelmingly want rules to protect them from technology’s risks, even if they don’t agree on who should be in charge.

• Support for guardrails is broad and bipartisan. Across Democrats, independents, and Republicans, strong majorities of voters support protecting kids, developing a comprehensive AI plan, and establishing national standards. 
• Voters have little appetite for a hands-off approach. Just 5% say AI doesn’t need regulation. A proposal to eliminate state laws that slow down innovation only drew 15% support—the lowest support rate of any of the proposals we polled. 
• But there is no consensus on who should regulate technology. Fifty-one percent of voters combined think either the federal government (30%) or state and local governments (21%) are best suited to regulate AI and other tech issues, but neither institution comes away with a strong mandate. 

• Voters don’t have much trust in either political party. Both parties are underwater: 34% of voters view Democrats favorably and 53% unfavorably, while 37% view Republicans favorably and 52% unfavorably.
• Voters also don’t trust tech companies to police themselves. Social media companies scored even worse on favorability than the political parties (20% favorable/58% unfavorable). Voters rejected the idea that tech companies should come up with child-safety practices on their own (7%).

6. Voters Want Strong Guardrails Without Giving Up the Benefits of AI.

Voters have very real concerns about AI, but they aren’t ready to give up on its potential. They want policymakers to put strong protections in place, even if that slows innovation. But when given a choice between competing approaches, voters favor a middle path: manage the risks and promote the benefits. 

• Voters see real upside from AI. Voters see tangible benefits, particularly when it comes to scientific and medical research. Eighty-one percent think it is likely AI will speed up scientific research and unlock breakthroughs in medicine.
• The most popular approaches pair protection with progress. Eighty percent agree the country needs a comprehensive plan to understand AI, protect people from risks, and carefully implement the technology. And 76% agree that we need national standards to protect against harm while maintaining US leadership in AI. 

• Voters reject the extremes. Approaches that prioritize innovation above all else or seek to dramatically slow AI development are less appealing than strategies that balance the two. Proposals to pause data center construction (65% support) and eliminate state-level laws that slow down AI (37% support) landed in the bottom three policy preferences.
• The bottom line: voters want policymakers to manage AI, not stop it. They are willing to accept some tradeoffs in speed and innovation for stronger protections, but they don’t want to sacrifice AI’s potential benefits altogether.