AI-made political ads surged in the 2026 US midterms, researchers say

In short: AI-assisted political ads and viral fakes increased sharply in the 2026 midterms, but the bigger impact may be growing confusion about what to trust.

What's going on

Researchers tracking the 2026 U.S. midterm elections saw a big jump in political ads that were generated or edited with AI. About 300 ads were made or enhanced this way, and roughly $187 million was spent promoting them. That is a steep increase from about $50,000 during the 2024 election cycle.

The Wesleyan Media Project found that around 90% of the AI-linked ads they tracked came from Republican candidates or groups aligned with them. Some ads showed events or quotes that did not happen, while others used AI in quieter ways, like tweaking real video or audio. Researchers said it is hard to measure how much these ads changed votes, and they may have mostly strengthened opinions people already had.

Researchers had worried about highly convincing “deepfakes,” meaning realistic fake video or audio that makes someone look or sound like they said something they did not. That feared wave did not clearly appear in a way that changed election outcomes. Some of the biggest viral fakes were more silly than persuasive, like an AI-made video riffing on the TV show The Office that got over 16 million views on X.

What to watch

The longer-term issue may be trust, not who won. As more fake content spreads, real footage can be dismissed as fake, like a real receipt being called a forgery. Researchers also warned about chatbots giving wrong voting information, such as outdated rules or deadlines, which could mislead voters in practical ways.

Source: NYTimes