AI-Made Deepfake Political Ads Targeted by Michigan Lawmakers

Michigan lawmakers are advancing bipartisan legislation to crack down on AI-generated deepfakes in election materials in a state where voters could determine control of the White House and the U.S. House in 2024.

The legislative package could add Michigan to the handful of states restricting the use of artificial intelligence in political campaigns in response to worries about how the technology could empower bad actors to sway voters.

Bipartisan support is one advantage Michiganders have to get the restrictions approved in time for the 2024 election, Rep. Matthew Bierlein (R) said in an interview. Michigan’s “purple state propensity” makes it especially important for the parties to come together, he said. “We could all be either really helped or really hurt by AI-generated ads” with the White House, Congress, and state legislature on the line.

Michigan’s five-bill legislative package would require political messages to include disclosures of AI use, with repeat offenders facing felony convictions and two-year prison sentences. Anyone convicted of using deepfakes while attempting to sway an election could receive up to five years if the measures get enacted.

Individual candidates and the state attorney general would be able to sue over any political materials that appear to violate such rules.

The Michigan House Elections Committee approved the package Tuesday.

Committee Chair Penelope Tsernoglou (D) showed colleagues a presentation featuring bogus audio of President Joe Biden congratulating her for sponsoring two of the bills to demonstrate how easily deepfakes can be made.

She told Bloomberg Law she aims to get the full chamber to pass the bill before the end of the year. “It’s really important for this bill to pass as soon as possible because we already have this type of thing happening—and it’s only growing,” said Tsernoglou. December 21 is the last day the legislature is scheduled to meet this year.

Beyond FEC

Michigan lawmakers are looking to act in ways that won’t run afoul of federal law. The Federal Election Commission is considering expanding rules barring “fraudulently misrepresenting other candidates or political parties.” Such regulatory decisions typically take a few months.

The Michigan bill package aims to cover videos featuring phony voters and other made-up footage that doesn’t necessarily feature a candidate or political party that would be covered by the potential FEC change.

• HB 5141 would prohibit using AI in campaign materials without disclosure.
• HB 5142 would punish violators of H.B. 5141 by creating a Class G felony punishable for a maximum of two years.
• HB 5143 defines AI as “as a machine-based system that can make predictions, recommendations, or decisions influencing real or virtual environments for a given set of human-defined objectives.”
• HB 5144 would ban distributing “materially deceptive media” to sway an election.
• HB 5145 would create a Class E felony with a maximum of five years imprisonment for violating HB 5144.

Michigan would join California, Texas, and Minnesota in curbing deepfakes if the bill package passes the state Senate and gets signed into law by Gov. Gretchen Whitmer (D). New York lawmakers are considering similar legislation (AB 7106).

Spotting the Difference

Enforcing such laws might be a challenge considering how realistic AI-generated audio and video can appear, especially when it does not feature famous people like Donald Trump or Joe Biden, both of whom have been targets of deep fake hoaxes this year.

Experts can examine footage at “the pixel level or the video frame level” to verify authenticity, but “it’s really difficult” for ordinary people to spot the difference between a real human talking and a digital counterfeit, Gang Wang, an associate professor of computer science at the University of Illinois, Urbana-Champaign, said in an interview.

“People should be worried” about the AI’s potential to spread misinformation though it is not a problem that “we cannot handle,” said Wang.

Federal officials have acknowledged the looming danger of AI deep fakes manipulating voters, but they have yet to take definitive action to ban them or require their disclosure.

“As Americans prepare to go to the polls in 2024, we have to move quickly to establish safeguards to protect voters from AI-related misinformation,” said US Senate Majority Leader Chuck Schumer said in September. “ “It won’t be easy. For Congress to legislate on AI is for us to engage in perhaps the most complex subject this body has ever faced.”