What Latino Voters Can Teach Washington About AI Governance
Roberta Braga / Oct 29, 2026Roberta Braga is founder and executive director of the Digital Democracy Institute of the Americas.
A pedestrian passes a downtown mural honoring World War II veterans in California’s 22nd Congressional District on October 22, 2026 in Bakersfield, California, a competitive southern Central Valley district with a majority Latino population. (Photo by Mario Tama/Getty Images)
For years, the debate over AI regulation in the United States has been framed as a false dichotomy. Some argue if we move too slowly on innovation, the US risks falling “behind” its competitors. Others say if we regulate technology too aggressively, innovation suffers.
New polling of 1,500 US Latino registered voters by the Digital Democracy Institute of the Americas (DDIA) suggests Americans don't see those goals as mutually exclusive.
In particular, the highly consequential Latino electorate that tends to swing between the parties is looking for something more practical. Latinos want progress, but they also want rules that protect their jobs, personal information, and communities. This is the case for Latino Republicans, Independents, and Democrats.
Where are Latinos on AI and data centers?
Latino voters are adopting AI, led by young men between the ages of 18 and 29, but familiarity with the technology has not brought comfort. More Latino voters are worried than excited, and most expect AI to shape their lives.
82% of respondents believe that what happens with AI in the coming years will affect what happens in their life. Latino voters’ concerns about AI are mostly connected to perceptions about impacts on everyday life. 74% of respondents worry that AI will take jobs away from ordinary people, 72% worry about their personal data when they use AI, and 71% say it’s hard to tell what is and isn’t made by AI on the Internet. Fears about AI causing human extinction or a global catastrophe in the next few decades are less common, though still shared by 60% of people.
When it comes to data centers, our research found that 74% of Latino voters worry data centers will drive up electricity bills. Another 68% believe they put too much strain on local water supplies. By comparison, only 40% believe data centers bring good jobs and investment to the communities where they are located. Just 42% agree that building more data centers is necessary for the US to stay ahead of China and other countries in AI.
The gap between perceived costs and benefits is striking and becomes even clearer when the question turns local. Just 23% support building a data center in their own community, while 51% oppose it.
There are important partisan differences. Latino Republicans are more likely than Democrats to believe data centers bring jobs and investment, 55% to 32%, and that they are essential to US competitiveness, 58% to 32%. Yet even among Republicans, 72% want stricter rules governing data center construction and 64% worry about higher electricity bills.
This is where AI policy has an opportunity to move beyond the familiar innovation-versus-regulation divide. A serious infrastructure governance agenda should consider enforceable requirements for assessing electricity and water impacts, meaningful public participation in siting decisions, greater transparency around promised economic benefits, and safeguards against shifting infrastructure costs onto households.
It should also ask whether communities have the ability to negotiate the terms under which AI infrastructure is built, rather than simply being informed after the most consequential decisions have been made.
For Latino voters, AI worries start close to home
What, specifically, do Latinos want to see from policy?
One of the clearest findings from our research is that Latino voters overwhelmingly want stronger oversight of AI, regardless of political affiliation. Nearly eight in ten respondents (78%) believe there is a pressing need for stricter rules and oversight of AI. This includes 83% of Democrats, 73% of Independents, and 72% of Republicans.
Support extends to specific proposals. Two-thirds favor requiring companies to submit risk assessments to the US Department of Commerce before releasing new AI models. An equal share supports independent review and approval of AI systems before release. And 61% support pausing the development of more advanced AI until appropriate safety standards are established.
These findings challenge the idea that public support for technological progress requires a light regulatory touch. They also raise questions about the reliance on voluntary commitments and industry-led safety frameworks.
When asked whom they trust to keep people safe as AI develops, only 19% of Latino voters express trust in technology CEOs, and just 22% trust AI companies. Scientists are the only group tested that more respondents trust than distrust, with 50% expressing trust. Government institutions fare poorly, too. Only 27% trust Congress, and 28% trust the president to keep them safe as AI develops.
These very low levels of trust in elites and institutions create a difficult governance challenge. The public wants stronger oversight, but has little confidence in either the companies developing AI or the institutions responsible for regulating it. Trust must be earned through transparent, enforceable safeguards, independent scrutiny, and mechanisms that allow the public to hold both companies and regulators accountable.
What can policymakers take from these findings?
Days from the US midterms and as a future Congress prepares to take office, our poll findings provide a useful starting point for examining AI and online safety policy that serves Latino communities in a low trust environment.
First, policymakers should advocate for requiring meaningful risk assessments before powerful AI systems are released. 67% of Latino voters support having companies submit assessments to the US Department of Commerce before releasing new models, for example. Such assessments could examine how models enable deceptive content, impersonation, privacy violations, and risks to minors, including their potential use in misleading civic information.
Second, policymakers should consider requiring that AI companies have their systems reviewed by independent entities rather than relying exclusively on companies to evaluate themselves. 67% of Latinos we polled also support this approach. Reviews could assess the accuracy and reliability of AI systems across languages, including the widely spoken Spanish, and evaluate whether safeguards work in practice.
Third, policymakers should push to further improve transparency around AI-generated content. 75% of Latinos say it’s hard to tell what’s true and what’s false on social media these days and 71% agree it is hard to tell what’s AI-generated online. People want more tools to help them navigate pollution online. Practical standards for provenance, labeling, and source attribution could help people better understand where information comes from. Such measures should acknowledge that labels and watermarks have limitations and cannot independently establish whether a claim is true.
Fourth, policymakers must place children's safety at the center of AI product development. 83% of Latinos polled worry about the safety of children online. Age-appropriate design, stronger privacy protections, and independent testing of risks to younger users deserve consideration before products are widely deployed.
Finally, there is a broader lesson in these findings that point toward a different model of public engagement. Rather than consulting communities only after regulations are drafted or infrastructure decisions are nearly finalized, policymakers should incorporate public perspectives earlier and more systematically. That means investing in accessible public consultations, including Spanish-language engagement, supporting independent research into how different populations experience AI, and ensuring that community concerns have a meaningful pathway into regulatory decisions.
It also means recognizing that expertise is not confined to technology companies, academic laboratories, or government agencies. Workers, parents, educators, journalists, and local residents have knowledge about how these technologies operate in everyday life that should inform how they are governed.
The bottom line
The US is approaching a consequential period for AI governance. Per our data, neither political party in the US owns the issue of AI and data centers. Latino voters’ views on AI lean toward caution and regulation, positions where Democrats currently have an edge. But with about one in five saying they trust both parties nearly equally, and about one in eight saying they trust neither party, there is still an opening for either party to own the issue with this community.
As investment accelerates and the technology becomes more deeply embedded in everyday life, policymakers must make decisions about safety, infrastructure, economic competitiveness, and the distribution of benefits.
The temptation will be to frame these decisions as a choice between moving fast and falling behind. Our research suggests that this is the wrong choice to offer the public.
The next generation of AI governance should be judged by whether it establishes enforceable accountability, protects communities from foreseeable harms, and gives the public a meaningful role in shaping the technology's development.
Authors
