The Federal Election Commission is poised to take its first steps to regulate the use of artificial intelligence in political ads when it meets next week. It will vote on a proposed rule that would prohibit federal candidates and campaigns from making a “fraudulent misrepresentation” in political ads. The interpretative rule would make it clear that the false representations created by artificial intelligence are prohibited, while the FEC says it is part of much broader limits that apply irrespective of the technology used.
“This fraud may be accomplished using AI-assisted media, forged signatures, physically altered documents or media, false statements, or any other means. The statute, and the Commission’s implementing regulation, is technology neutral,” the draft rule says.
The FEC is going out of its way to say its regulation would cover more than AI-generated material, but Vice Chair Ellen Weintraub said at a House roundtable Wednesday that the interpretive ruling will make it clear that AI is covered.
“The interpretive rule could provide some immediate clarity, and I intend to support it as a transparency advocate,” Weintraub said. “The first essential step in regulating AI in elections would be to require greater transparency when it is used.”
FEC Chair Sean Cooksey in June asked the Federal Communications Commission to not “invade” his agency’s jurisdiction with its own plan to require politicians to label the use of AI in their radio and television ads. But FCC Chair Jessica Rosenworcel has continued to move forward with her proposal, telling lawmakers the FCC has legal authority over what broadcasters air and put in their public files.
“I understand the FEC is also considering a rulemaking on AI, and the more the merrier,” Rosenworcel said. “But with our complementary authority, the FEC can regulate AI use and online advertisements for federal candidates, while the FCC can focus on the areas where the FEC does not have power to act. For instance, the FEC does not oversee television or radio stations. Moreover, under the law, the FEC authority campaign is limited to federal political candidates. It doesn’t extend to independent issue campaigns for state and local elections. The FCC authority does, so we can address those loopholes.”
Because the FCC also regulates phone companies, Rosenworcel said it can also go after AI-generated robocalls. Such enforcement has already played out. It worked with the New Hampshire Attorney General to track down a man responsible for using AI-generated calls of President Biden just before the station’s January primary. “There’s just too much potential for AI to manipulate voices and images and political advertising to do nothing,” she added.
Across the country, there is a growing bipartisan consensus that efforts to rein in AI need to start somewhere. The federal efforts are playing catch-up to what is happening in the states. Nineteen states have already enacted laws to regulate the use of AI technology in elections.
“The risk of a deep fake October surprise has animated the public debate on AI, but there are more subtle risks as well, such as the risk of AI-enhanced micro-target spreading of disinformation at more speed,” said Weintraub. She said the FEC has received tens of thousands of comments on the use of AI in political ads, and the overwhelming consensus is they need to act.
Commissioner Brendan Carr has been critical of the FCC’s proposal addressing AI use in political radio and TV commercials (MB Docket No. 24-211). He has called the idea a “recipe for chaos,” and believes it is at odds with federal law that says stations cannot refuse the sale of airtime to federal candidates, even if the spot defames other candidates.
But Rosenworcel told lawmakers the First Amendment looms large over the FCC’s work, and requiring labels is a “simple place” to start.
“We oversee television and radio, and because we license them, if they fail to do this, we have enforcement authority to go after them, so there is a burden on them to engage in that kind of fact finding and make sure those filings are accurate,” Rosenworcel said. “I think that tool is powerful.”
Comments in the FCC proceeding are due Sept. 19.
Ultimately, it may take a change of federal law to address AI’s use in political ads. Rep. Joe Morelle (D-NY), the ranking Democrat on the House Administration Committee, said generative AI has “become embedded” and disclosures in the meantime will help “policing to some degree” its use. “It seems in this era that simple disclosure, is maybe not that simple,” he said.
コメント