Minnesota banned the use of deepfakes in elections. Now dozens of other states are, too.

Minnesota banned the use of deepfakes in elections. Now dozens of other states are, too.

Minnesota banned the use of deepfakes in elections. Now dozens of other states are, too.

In a nearly unanimous vote last spring, lawmakers in Minnesota made it a crime to use deepfake technology to influence an election.

The state became one of the first to ban the use of A.I.-generated material in elections, which officials worry could be used to spread disinformation about candidates or the voting process.

“We wanted to get ahead of the curve,” said Sen. Erin Maye Quade (DFL-Apple Valley), who sponsored the legislation.

Now, dozens of other states are following in Minnesota’s footsteps.

In the last two months, more than 30 states filed proposed legislation that would regulate or ban the use of deepfakes in elections, according to data reviewed by 5 INVESTIGATES.

The proposals vary widely. Some states are seeking to make it a crime, as Minnesota did. Others would only create a civil penalty. Some bills allow for disclaimers on materials, while others do not.

Regardless of what’s in the proposed legislation, some cybersecurity experts are skeptical these laws will tackle the problem.

“I think it’s going to be very difficult because it’s the internet,” said Hany Farid, professor of computer science at the University of California Berkeley.

Forty-two states

Six states – including Minnesota – already have laws that ban the use of deepfakes to include an election. A legislation tracker compiled by Public Citizen shows 36 others with pending legislative proposals as of March 6th.

Maye Quade said she believes the wave of legislation is a sign that officials are concerned about the rise of artificial intelligence programs, and their impact on society.

"It's important that when we make laws around elections, we do it proactively and not reactively," she said in a recent interview.

Deepfakes are already creating challenges around conducting elections.

Days before New Hampshire's presidential primary, investigators say voters in the state received a robocall with the voice of President Biden, encouraging people not to vote.

"It's important that you save your vote for the November election," the recording said in part.

New Hampshire's Department of Justice identified the audio as an A.I.-generated deepfake and is working to trace its origins.

If that had happened in Minnesota, Maye Quade said prosecutors could pursue a case against whoever created and disseminated the audio.

"Depending on how that was created, and disseminated, that could be a violation of this law," she said.

Skepticism

While states work to pass these proposals, cybersecurity experts and some lawmakers question the effectiveness of the laws.

Hany Farid, a computer science professor at the University of California, Berkeley, worked on California's first deepfake ban, which was enacted in 2019.

Farid said there are flaws in the practical application of these laws.

"If the actor is outside of your state, you have very little recourse at this point assuming that you can actually find them, by the way," he said. "We all know that the Internet is exceedingly anonymous."

In the foreground, an image of Hany Farid, expert in detecting deepfakes, in a laptop that's propped on a table. reporter sits in the background, taking notes, listening to Farid during the interview
Hany Farid, professor at the University of California, Berkeley, specializes in detecting deepfakes.

In 2022, lawmakers in California reauthorized its deepfake ban, which was set to expire in 2023. Research presented during the legislative session revealed that there is "scant evidence" that the law was effective at deterring the use of deepfakes from including elections.

In Minnesota, state court data obtained by 5 INVESTIGATES shows no one has been charged under the state's deepfake ban since it went into effect last May.

When presented with the data, Maye Quade acknowledged that prosecutions would be "difficult," but defended the statute as a tool in the toolbox.

"We wanted to at least have that ability should a situation arise where someone could utilize the statutes," she said.

The future

Maye Quade is looking to sharpen that tool by introducing updated language to the law that would disqualify a candidate for running for or holding office if they're found to have violated the deepfake ban.

During a committee hearing in February, several of her colleagues pushed back against the additions, arguing the law has yet to be meaningfully tested.

"We haven't had an election cycle yet since we passed the last law," said Sen. Mark Koran, a Republican who represents North Branch.

When asked about those criticisms, Maye Quade argued the additions are "clarifications" to the law.

"I want this to be proactive, so we have strong rules before we get to election, before we have a problem," she said.

state senator erin maye quade, left, sits across from 5 eyewitness news reporter kirsten swanson during a recent interview at the state capitol
Sen. Erin Maye Quade (DFL-Apple Valley), left, speaks with 5 EYEWITNESS NEWS during a recent interview.