Half of U.S. states seek to crack down on AI in elections
Add Axios as your preferred source to
see more of our stories on Google.
As the 2024 election cycle ramps up, at least 26 states have passed or are considering bills regulating the use of generative AI in election-related communications, a new analysis by Axios shows.
Why it matters: The review lays bare a messy patchwork of rules around the use of genAI in politics, as experts increasingly sound the alarm on the evolving technology's power to sway or disenfranchise voters.
- There have already been instances of generative AI being "used to confuse — and even suppress — voters," Sen. Mark Warner (D-Va.) told Axios in an email.
- "I don't think genAI developers or platforms are taking the misuse potential serious enough," added Warner, chair of the Senate Intelligence Committee.
Catch up quick: In January, a spate of fake robocalls in New Hampshire used an AI-generated impersonation of Biden's voice to urge Democrats not to vote in the state's primary.
- In another high-profile incident, last July a super PAC backing Florida Gov. Ron DeSantis in the GOP primary used AI to imitate Trump's voice in an ad attacking him. The Trump campaign slammed the ad as a "desperate attempt" to "deceive the American public."
The big picture: There are few federal guardrails in place to regulate the use of AI — even President Biden's AI executive order is largely voluntary, with little enforcement power.
- Senators have introduced two bills to regulate genAI in election campaigns, but they've yet to pass — even as many Americans fear the technology will hurt elections.
Which states have enacted laws
Nineteen states have already passed laws to address deepfakes — artificial media made with AI to show people or events that aren't real — and AI-generated "synthetic media" in elections.
- In 2019, Texas became the first state to ban the creation and distribution of deepfake videos intended to hurt a candidate or influence an election.
- That same year, California passed a law banning the distribution of "materially deceptive audio or visual media" meant to damage a candidate's reputation or influence a voter within 60 days of an election.
In 2023, Minnesota and Michigan banned the distribution of AI-generated deepfakes within 90 days of an election.
- Washington also passed a law last year targeting "synthetic media" that shows "realistic but false" depictions of candidates. Candidates depicted in the AI-generated material can sue for damages unless the media includes a disclosure.
New Mexico, Florida, Utah, Indiana and Wisconsin have enacted laws this year requiring disclosure when AI-generated content is used in campaign communications and political ads.
Which states are considering laws
At least seven states are considering laws to crack down on the use of AI when it comes to elections.
- Pennsylvania, Massachusetts and North Carolina are considering bills requiring disclosure statements on AI-generated or AI-altered media depictions of a candidate.
- Delaware lawmakers have passed a bill cracking down on the distribution of deepfakes to influence an election, but the measure is still awaiting the governor's signature.
Caveat: A slew of states — including Alaska, Oklahoma and Louisiana — introduced bills this year to regulate AI in election communications. All efforts died in committee or were vetoed.
Between the lines: Some laws passed or being considered provide exceptions, such as if the media includes a disclosure, or if a political ad is satire or parody.
- States also vary regarding classifying violations as criminal acts or misdemeanors, and if victims may sue for damages.
What experts are saying
AI-generated content does offer some benefits for political campaigns, Alix Fraser, director of the nonprofit Issue One's Council for Responsible Social Media, said.
- But deepfakes can be a problem, and many state proposals have limited time frames. Major races play out over a much longer cycle than just a few months, Fraser told Axios.
- While focusing on political ads is important, a much more comprehensive framework — such as flexible laws and a federal digital agency that could act and adapt as AI technology evolves — is needed, Fraser said.
The bottom line: Despite growing activity on the state level, "at the end of the day, it's only going to be piecemeal, it's never going to be sufficient, until we have federal action that covers the entire country," Fraser said.
Go deeper:
