Policymakers and legislatures work to contain the harm AI and deepfakes could do in this year's elections

(Graphic by Sean Westwood/DALL-E artificial intelligence
via Dartmouth)

AI and deepfake political ads are easy to disseminate and pose a threat to U.S. elections. Policymakers have been hard at work trying to get ahead of the problem. "The Federal Communications Commission announced it would move forward with a proposal to require that TV and radio advertisements disclose on air when AI is used," reports Chris Teale of Route Fifty. "The timeline for the rule’s implementation is unclear, but it is expected to go into effect by November."

The proposal's green light wasn't unanimous. Teale writes, "Commissioner Brendan Carr said in a statement that the FCC’s proposal, which will be subject to a vote later this year, is a 'recipe for chaos' that can 'only muddy the waters.'. . . 'Suddenly, Americans will see disclosures for AI-generated content on some screens but not others, for some political ads but not others, with no context for why they see these disclosures or which part of the political advertisement contains AI."

One deepfake trick is to generate robocalls to voters. Before the "New Hampshire presidential primary election, a robocall purportedly featuring President Joe Biden discouraged voters from going to the polls, and instead told them to wait until November’s general election to vote," Teale explains. "The call turned out to be a deepfake generated by artificial intelligence."

The FCC's rule is "one of many efforts in recent weeks to get ahead of the potential threat deepfakes pose. State and local election offices have already seen how AI-generated content could be used to spread misinformation and disinformation," Teale reports. "President Joe Biden has already been the subject of much AI-generated content, including a video on Facebook that made it look like he inappropriately touched his granddaughter’s chest after they voted in the 2022 midterms."

AI leaves voters unsure of who's saying what and possibly defaulting to believing nothing anyone says. In response, policymakers and legislatures are working through the best ways to help create clarity where there could be chaos. Teale writes, "Already, several state legislatures have passed laws requiring disclosure of AI’s use in elections and campaigns, with only Minnesota and Texas outright banning it. More than three dozen states are in various stages of considering similar disclosure bills."