AI's Impact on US Elections Falls Flat

The technology, or social media, was actually in the forefront of the recent elections. In 2020, America followed every decision made by platforms: whether to remove Trump’s election fraud posts, label them as warnings, or temporarily ban their author. Social media platforms were at the center of the election story: from Russian influence and Cambridge Analytica to discussions about whether the president should be banned from platforms. Every change in rules was news, and every moderation decision was a political event. Later, platforms gave up on strict moderation and quietly withdrew from their role as election guardians, and no one claimed to have decided the outcome. This year, many expected a strong impact of AI on the final result, but surveys show that voters barely think about it.

Voters don’t know who to trust with AI regulation. In a September survey by The New York Times and Siena, only 1% of likely voters mentioned AI (or data centers) as the most important issue for their vote. The economy dominated, especially with fuel prices and living costs that America hasn’t seen in a long time. A similar survey by the industrial lobby NetChoice showed that 55% of likely voters put inflation and living costs as top priorities, and AI regulation only 9%. Voters don’t even know who to trust with AI regulation: 42% trust Republicans, 40% trust Democrats, which is practically tied and follows party divisions.

Bots wrote election guides for other bots. The fear of deepfakes didn’t materialize in the form predicted. The president and federal agencies regularly post AI-generated images, and outrage rarely lasts more than a day, after which everyone waves their hand. AI isn’t in the first election news, but it has found its way to voters. The Washington Post tested chatbots on election issues for all 3,100 US counties and found that they often rely on three small, newly launched pages that are almost entirely created with AI agents. Microsoft’s Copilot cited one of them, Ballot Brief, in 75% of responses, and ChatGPT in 22%.