Explainer: How AI muddies the waters ahead of US midterm elections?

AI-driven disinformation is flooding voters' feeds as the United States heads toward its Nov. 3 midterm elections. [IstockPhotos]

AI-driven disinformation is flooding voters' feeds as the United States heads toward its Nov. 3 midterm elections.

"AI is making its way into the 2026 midterms in a big way this year," said the Wesleyan Media Project, which tracks AI-generated political advertising. 

Earlier in October, the project said it had identified 220 such advertisements worth more than 137 million US dollars, a staggering increase from 50,000 dollars during the 2024 election cycle, and an "almost certainly" low estimate given the difficulty in detecting the use of AI.

With much of this content carrying no disclosure that AI had been used, analysts warned that it could mislead voters and undermine the integrity of elections.

AI ads and lies

Some AI-generated advertisements have been particularly provocative, such as one showing a Republican candidate for governor embracing Hillary Clinton, cheek to cheek and hand in hand, beneath the slogan "Clinton Lover. Trump Hater."

Others have gone even further. One depicted Democratic Senate candidate Abdul El-Sayed laughing at cards that read "Death to America" and "Celebrate 9/11."

Another showed James Fishback, a Republican primary candidate for Florida governor, hosting a party with Iranian leaders who used scimitars to attack the Star of David, a recognized symbol of Jewish identity and the religion of Judaism.

"You can run an information operation now with quality that was impossible in 2016 for super, super cheap," said Frances Haugen, a former Facebook data engineer turned whistleblower and advocate for transparency and accountability in big tech.

Those seeking to spread disinformation can get around content moderation rules "because you can just keep tweaking the post a little bit each time you create them, and some of them are going to get through," she said.

Meanwhile, a survey by the American Association of Political Consultants Foundation found that though only 1 per cent of respondents considered AI their most credible source of election information, one in five voters plans to use AI to look into a race on their ballot.

Twice that proportion has already received election details through an AI search summary, according to the survey. 

A cottage industry has quickly emerged, advising candidates on how to shape what AI says about them.

The left-wing advocacy group Run for Something has launched CampSight, a platform that tracks what large language models say about issues or candidates and recommends ways to shape those narratives.

On the right, the Center for Campaign Innovation (CCI) has published research on how AI models respond to questions about candidates and which sources they cite most frequently.

"We're campaigning for the AI vote now," Eric Wilson, executive director of the CCI, told The Atlantic, an American magazine and multi-platform publisher.

 

Lobbying against regulation

The AI giants, however, are pouring millions of dollars into lobbying to block unfavorable legislation and shape the political landscape ahead of the midterm elections.

The Wall Street Journal reported last year that Silicon Valley was channeling more than 100 million dollars into a network of political-action committees (PACs) and advocacy groups to fend off stringent AI regulation and defend the industry's interests in the midterm elections.

Among the biggest efforts is Leading the Future, a super PAC backed by OpenAI President Greg Brockman and venture capital firm Andreessen Horowitz.

The group reportedly spent 8 million dollars seeking to unseat Democratic New York State Assemblymember Alex Bores in a congressional primary, targeting him over his sponsorship of the RAISE Act, an AI safety bill signed into law last December.

It came amid Washington's continued efforts to stop or limit state-level AI legislation.

In 2025, a group of Republican senators proposed a moratorium on state AI laws as part of budget reconciliation, following a last-minute lobbying push by tech groups eager to roll back rules they saw as barriers to unfettered innovation.

In December, U.S. President Donald Trump issued an executive order establishing an AI litigation task force to challenge state laws deemed more than "minimally burdensome" and directing the Commerce Department to consider withholding federal broadband funds from states with "onerous" AI regulations. [IstockPhotos]

Though the Senate rejected that proposal, related preemption efforts continued to emerge in other forms.

In December, U.S. President Donald Trump issued an executive order establishing an AI litigation task force to challenge state laws deemed more than "minimally burdensome" and directing the Commerce Department to consider withholding federal broadband funds from states with "onerous" AI regulations.

Disinformed voters

Analysts warned that undisclosed AI advertisements can manufacture alternate facts, making it harder for voters to distinguish reality from falsehoods.

Susan Hyde, a professor of Political Science at UC Berkeley, said the deeper concern is whether voters could make informed decisions when they could no longer trust the information they receive.

"Classical theorists of democracy focus on the necessity for citizens to be informed about their choices in order to make an educated choice about who to vote for," Hyde said, adding, "The other thing that's interesting is the ease with which politicians can use propaganda-style communication and lie to their supporters directly."

Similarly, the Brennan Center for Justice said transparency and human oversight are lagging behind AI's rapid development, adding that its research offers a glimpse of how deepfakes, readily enabled by chatbots, could become "potent tools for electoral cheaters."

It found that chatbots like ChatGPT, Gemini and Grok failed to decline requests to generate images playing on familiar election conspiracies, such as election workers stuffing ballot boxes or large numbers of non-U.S. citizens voting.

"We can't have a robust debate or fair electoral decisions if we cannot trust the information by which we vote and with which elections are run," the center said.

"Policymakers need to act quickly to find ways to curb the bots' tendency to create or ignore election disinformation."