Three Things the Midterm Election Tells Us About AI and the Likely Future of AI Policy

The 2026 midterm elections are significant for the ongoing debate over AI policy in three ways: first, the role that AI played in the election itself; second, the outcomes of AI questions on the ballot and AI’s role in voter decision-making; and third, the impact of election results on existing AI policy debates. 

AI Disinformation Still Isn’t an Election-Disrupting Problem 

With AI’s rising popularity, the midterms are not the first time we’ve seen concerns over AI-powered election disruption. The 2024 election saw a great deal of rhetoric around such potential concerns. In fact, 2024 saw the World Economic Forum label AI-powered misinformation the world’s greatest short-term threat. But the worst-case scenario did not come to pass in 2024. In fact, despite these concerns, AI tools in the 2024 election had an underwhelming impact. 

Two years later, AI and its applications have advanced. Unsurprisingly, fears of AI misinformation and disinformation in elections reemerged as the midterms neared. In the immediate lead-up to the midterm election, there were headlines about leading AI companies forming “war rooms” to fight election disinformation and monitor for possible election interference. But once again, it seems that AI disinformation has not become a direct problem. The reality is voters and other internet users are quite discerning with the information they receive. 

As my colleague David Inserra has written in the past, “The prevalence and impact of online misinformation is vastly misunderstood and overstated. Meaningful misinformation is still a relatively small amount of content online and is concentrated in communities already predisposed to believe it.” This seems to be playing out in the AI context, as it has in the case of any number of media and technologies that preceded it. 

Of course, it is important to be aware of the information ecosystem. Providing tools that can help identify potential AI use can be useful. As can information from platforms that can help users understand context or potential involvement, such as community notes, the ability to turn on AI labels, and information on potential foreign involvement. As AI tools continue to evolve, the answer is not to panic or require government to serve as an arbiter of truth, but to look to the market and the various ways platforms, civil society, and users provide the education and tools to help voters navigate the information. 

Was AI on the Ballot? 

The closest AI came to being directly on the ballot was a number of ballot measures on data centers. But there were also questions in some key races about whether AI might be a motivating factor for voters. 

Many who oppose AI or data centers have trumpeted that, when on the ballot, opposition to data centers and measures in favor of moratoriums won the vast majority of the time. But this does not tell the full story of what actually happened. Twenty-one of the ballot measures that passed to restrict data centers were in Ohio. The nature of these restrictions also varied, from more explicit city and county bans or moratoriums to regulations that require approval or address zoning issues around data centers. 

While I have written about the importance of data centers as infrastructure in the past, these decisions are best made by local communities—including when they reject them. However, it does raise a few concerns. First, unfortunately, some like Senator Bernie Sanders have taken these rejections as a reason to call for a national ban. As I have written in the past, such bans would have a problematic impact not only on innovation but also raise concerns about federal intervention into local land use issues more generally. The concerning anti-data center momentum has consequences for communities beyond those where it was on the ballot. For example, Elon Musk has once again floated the idea of data centers in space, while Google announced a $15 billion expansion of data centers in Malaysia. These trends may mean local US communities lose out on the economic benefits of data centers and that US innovators and businesses, particularly smaller and mid-size ones, may find themselves encountering concerns about the sufficient capacity for the AI and cloud computing tools they want to implement. 

But did AI play a role in voters’ overall approach to the election? 

The answer is maybe. AI was on many voters’ minds, but it was not a defining issue for many voters. For example, an October NYT/​Sienna poll found that less than 1% of the electorate considered AI and data centers to be the most important issue in deciding their vote for Congress. A recent Gallup poll found that voters ranked AI and data centers 10th out of 11 issues, ahead of only climate change. Americans want to see a thriving economy above all else. While AI was a factor, it ranked last in several NBC exit polls behind issues like cost of living and corruption. As with most recent elections, voter concerns today are dominated by the economy. 

This does not mean that AI policy wasn’t a factor in the decisions of many voters, particularly in certain key races. For example, 55% of Ohio voters labeled AI and data centers a crucial or major factor in their vote for Congress. This is similarly reflected by 57% in Texas and 53% in Mississippi. On the issue of whether AI will hurt society, 58% agreed in Ohio, 56% agreed in Texas, and 55% agreed in Mississippi. These numbers may indicate a strong and concerning national public sentiment that is not concentrated in one portion of the electorate. Additionally, voters who stated AI and data centers were a crucial or major factor were more likely to have favored the democratic candidate in those same races. 

What Does the Election Mean for AI Policy? 

The “Blue Tsunami” means that Democrats will control both chambers of Congress. Some have already called for AI hearings or restrictions. Yet, these broad regulatory calls would encounter friction with a White House that has emphasized the importance of a light-touch, self-regulatory approach to allow American companies the opportunity to be the global leader in this technology. (Of course, there are notable exceptions to this approach, for example, last summer’s calls for government investment in leading AI companies.) 

In divided government, there could still be opportunities on specific issues even in the absence of a federal framework. Many of the current concerns about AI relate to privacy, emphasizing once again the need to consider a potential federal data privacy framework. Similarly, there might be opportunities for bipartisan frameworks on particular applications or concerns. Some of these might be positive, like guardrails on the underlying issues of surveillance and autonomous lethal weapons at the heart of the Pentagon-Anthropic dispute. In other cases, such as potential restrictions on chatbots and youth online safety, bipartisanship and the momentum to “do something” could lead to proposals that still raise serious free expression and privacy concerns and fail to consider the negative impact on the people they claim to benefit. 

If debate on AI issues is likely to stall at the federal level, there could be an influx of state action. The last two state legislative sessions have already seen over 1,000 AI-related bills introduced. While these proposals vary significantly, many tackle issues that will have an interstate impact. The problem is that AI issues like safety and model development are inherently going to impact innovation far beyond a single state’s borders, resulting in an effect where a law from Denver, Sacramento, Tallahassee, or Albany could become de facto federal regulation. Even laws that may initially seem more targeted, such as bans on AI discrimination, can have a broad impact on access to information or expressive rights in design choices that reach far beyond a single state’s borders. 

For state legislatures looking to address intrastate issues related to AI, there are a few opportunities that could have a positive impact. For example, states could take steps to ensure that civil liberties are protected when it comes to concerns about potential abuse of AI tools in a state-local context. State policymakers could consider whether there are opportunities for changes to existing burdens in state regulations, such as occupational licensing that prevents potentially beneficial uses of AI or needs guidance to allow the development of important AI tools. 

In both Congress and state legislatures, the conversation around AI policy is likely to be ongoing. As policymakers consider what, if anything, to do, they should not only consider how to respond to any specific concerns but also ensure we are not preventing American innovators from pursuing beneficial uses of AI that we could not even imagine in 2026.