Have Politicians Changed Their AI Stances for the Midterms?

In an electorate as polarized as ours, it's rare to watch an issue form in real time. Most debates arrive pre-sorted: you can usually guess where a candidate stands on taxes, immigration, or abortion before they open their mouth. But every so often, something new breaks through, something that hasn't yet been claimed by either party. When that happens, what do politicians do? Do they lead, or do they listen?

Artificial intelligence is one of those rare issues. It's a fast-moving technology that promises to reshape the workforce, redirect how we allocate energy and water, and, depending on who you ask, possibly threaten humanity itself. It's developing faster than most people can track, and American opinion on it hasn't hardened along the familiar Republican and Democratic lines. That makes it a perfect test case for a simple question: when the public starts talking about something new, how quickly do the people asking for their votes start talking back?

How’d We Do It

To find out, we built a panel of midterm candidates most likely to be paying attention. We gathered the social media handles of every candidate running in races that Cook Political Report rates as Lean Democrat, Toss Up, or Lean Republican as of September 23, 2026. Our reasoning was straightforward: candidates in competitive races have the most to gain from reading the room. They're more responsive to current conversations, and more likely to post about where they stand. If any group of politicians was going to react to shifting public sentiment on AI, it would be this one. Using Infegy Starscape, we compared the Midterm Politician Panel and the over US AI conversations.

A Conversation That Keeps Getting Louder

Over the past year, public discussion of AI and data centers has grown quickly, but not in a straight line. As Figure 1 shows, the conversation has moved in waves.

Figure 1: US Conversation Volume of AI (September 24, 2025 - September 23, 2026); Infegy Social Dataset.

Figure 2: Midterm Candidate Panel AI Conversation (September 24, 2025 - September 23, 2026

The first peak came the week of February 23, driven by news that Anthropic had been removed as a Department of War (Defense) contractor. The second arrived the week of May 11, when debate over data centers, particularly in Utah, hit a high point. And since August, the conversation has surged dramatically as the idea of "p(doom)," the probability that AI leads to catastrophe, moved from niche forums into mainstream media.

So how did our panel of politicians respond to each wave?

February: Thinking Locally

The February spike was a national story. Anthropic's removal as a Department of War (Defense) contractor put AI, national security, and the relationship between tech companies and the government at the center of the conversation. You might expect candidates in competitive races to jump on a headline like that. They didn't.

Instead, the only AI-related discussion we saw from our panel was about data centers in their own districts, with some candidates in favor and others opposed. Even as one of the biggest AI stories of the year played out in Washington, competitive-race candidates kept their focus on the projects breaking ground in their own backyards.

Figure 3: Example post by Michigan Democratic Senate Candidate Abdul El-Sayed (March 1, 2026); X/Twitter.

May: The Data Center Debate Takes Shape

By the May 11 peak, candidates were starting to engage more directly. We recorded 9 responses from politicians in our panel. Some emphasized the importance of data centers for economic growth, while others made clear they didn't want them. Interestingly, the partisan split wasn't clean. Republican candidates landed on both sides of the issue, while the Democrats who weighed in were opposed. This is exactly what you'd expect from an issue that hasn't yet been sorted: the parties haven't handed their candidates a script, so they're improvising.

Summer: Politicians Get Ahead of the Public

Then something unusual happened. Starting in June, candidates began talking about data centers more and more, and they kept increasing that attention through the summer, even though public conversation wasn't growing at the same pace. (Figure 2)

That gap is telling. If politicians were simply following the national mood, their posting should have tracked the public's. Instead, it outpaced it. That suggests one of two things: either candidates were responding to intensely local concerns in their districts that don't show up in national volume, or they were being pressed on the issue by reporters more often than the broader public was raising it themselves. You can see this with how Premium News providers kept the momentum in the summer in Figure 4. Either way, data centers had become something candidates felt they needed to have a position on.

Figure 4: Premium news conversations discussing AI (September 24, 2025-September 23, 2026); Infegy Premium News.

August Onward: A Turn Toward Skepticism

Since August 18, as the broader conversation about AI has accelerated sharply, politicians have been more receptive than ever, and their tone has shifted. The opinions they're sharing have grown noticeably more negative.

This is the clearest sign yet that candidates are paying attention. The public mood moved, and the politicians moved with it.

Figure 5: Positive Percentage - Negative Percentage of both Midterm Politician Panel and overall US conversations about AI (September 25, 2026 - September 24, 2026); Infegy Social Dataset.

Converging on the Mean

The numbers tell the story most clearly. Over the past year, American sentiment toward AI has fallen from +23 points positive to just +4. (Figure 5) Politicians started from a far more enthusiastic place. In some weeks, their posts were a full 100 points net positive. Today, they sit at around +15.

That means candidates are still more optimistic about AI than the average American. But the trend is unmistakable: they've been regressing toward the public mean, steadily closing the gap between their messaging and the mood of the voters they're courting.

Listening, But Enough?

What this shows is that when Americans talk about AI, politicians are listening. Candidates in competitive races have followed the public from local data center debates to broader anxieties about the technology's future, and their enthusiasm has cooled as the public has.

The question heading into the final stretch of the midterms is whether they're listening closely enough. Public sentiment is still sliding, and candidates remain more bullish than the people they want to represent. Whether they close that gap, or whether voters notice that they haven't, is something we'll be watching closely in the coming month.

Want to know more about Infegy Starscape? Request a demo today.