An announcement by the U.S. Federal Election Commission on Thursday that it will not take action to regulate artificial intelligence-generated "deepfakes" in political ads before the November elections amounted to "a shameful abrogation of its responsibilities," said a leading critic of the technology.
A year after consumer advocate Public Citizen filed a petition with the FEC to demand rulemaking that would prohibit a political candidate or advocacy group from misrepresenting political opponents using deliberately deceptive deepfakes—fake images generated with AI—FEC Chair Sean Cooksey told Axios the commission will not propose any new rules this year.
Cooksey, a Republican, said he plans to close the pending petition on Thursday without taking any action, telling Axios that rulemaking to limit or prohibit AI in campaign ads would "overstep the commission's limited legal authority to regulate political advertisements."
"The better approach is for the FEC to wait for direction from Congress and to study how AI is actually used on the ground before considering any new rules," said Cooksey.
In other words, said Robert Weissman, co-president of Public Citizen, the FEC will "wait for deceptive fraud to occur and study its consequences before acting to prevent the fraud."
Weissman pointed out that while social media companies have made some rules to prevent political ads with AI from being posted, X owner Elon Musk himself recently posted a deepfake video on the platform that manipulated an image of Democratic presidential nominee and Vice President Kamala Harris, making it appear as though she was saying she was the "the ultimate diversity hire."
Musk posted the video in violation of his own company's rules, proving that "platforms cannot be trusted to self-regulate," Weissman said.
"Political deepfakes are rushing at us, threatening to disrupt electoral integrity. They have been used widely around the world and are starting to surface in the United States," added Weissman. "Requiring that political deepfakes be labeled doesn't favor any political party or candidate. It simply protects voters from fraud and chaos."
Weissman recently said on a newscast that without a ban on deepfakes in political ads, "it's entirely possible that we're going to have late-breaking deepfakes before Election Day, that show a candidate drunk or or saying something racist or behaving in an outrageous way, when they never did any of those things."
Weissman pushed back on Cooksey's claim that regulating deepfakes is out of the commission's realm.
"The FEC is the nation's election protection agency and it has authority to regulate deepfakes as part of its existing authority to prohibit fraudulent misrepresentations," said Weissman. "It should have acted on this issue long ago, before Public Citizen petitioned for rulemaking. When we did petition, the agency should have promptly acted to put a rule in place. It still could and should reverse the wrongheaded decision that Chair Cooksey has said is imminent, and act to protect voters and our elections."
Twenty state legislatures have taken action to prevent deepfakes from flooding local airwaves as voters prepare to head to the polls in the fall, but Weissman said the FEC's refusal to act "underscores the need for congressional action" and for the Federal Communications Commission to move forward with its own AI proposal.
The FCC in May proposed rules requiring on-air and written disclosures in broadcasters' political files when political ads contain AI-generated content.
