'Easily sway a close election': Experts warn AI 'deepfakes' pose real threat to democracy

With January approaching, lawmakers are preparing to gather at their respective state capitols for their annual legislative session. Tech experts are urging them to address the growing threat of "deepfakes" generated by artificial intelligence prior to the 2024 election.

"Deepfakes," in which generative AI creates an image or video of a public figure saying or doing something out of character, have become increasingly more common in recent years. According to NBC News, experts and advocates are warning lawmakers that they could soon make it virtually impossible for voters to determine the veracity of what they're seeing online.

"A deepfake released shortly before Election Day — perhaps showing a candidate drunk, or speaking incoherently, or consorting with a disreputable figure — could easily sway a close election," Robert Weissman, president of watchdog group Public Citizen, told NBC News.

"A torrent of deepfakes could leave voters unable to distinguish what’s real from what’s synthetic."

An April article from tech publication Wired warned that 2024 could very well be the first "deepfake election" as AI becomes increasingly more ubiquitous. Already, a super PAC allied with Ron DeSantis produced an attack ad using AI to generate fake audio of former President Donald Trump supposedly attacking Iowa Republican Governor Kim Reynolds. And because federal legislation aimed at regulating AI has failed to move forward, experts are now calling on states to pass their own bills to curb deepfakes' effects on elections.

"It’s certainly the case that the states unquestionably need to do more," Daniel Weiner, director of the elections and government program at the nonpartisan Brennan Center, told NBC. "I don’t think we can afford to wait."

NBC News pointed to efforts in Michigan, Minnesota and Washington state as examples of how states have attempted to address AI deepfakes. In Minnesota, lawmakers passed a bill in August that bans the publication of "deepfake media to influence an election" 90 days before Minnesota voters go to the polls in any state election.

A bill passed last month in Michigan has a similar 90-day ban. However, the bill stipulates that deepfake producers can get around the ban if they clearly label that their content contains "manipulated" video, audio, text or images. And in Washington state, legislation passed in May requires that any media containing "synthetic" images, audio or video must be clearly labeled.

Robert Weissman stressed that AI regulation should come sooner rather than later as deepfakes become more common. He added that AI has a poisonous effect on media literacy as a whole, warning that "the prevalence of deepfakes could enable candidates to deny the validity of authentic content."

READ MORE: Viral screenshots of Trump supposedly blaming Biden for lemonade-related deaths are fake