With generative artificial intelligence (AI) proliferating across all industries as the general election gets underway, it should come as no surprise to see AI-generated political advertisements increasingly deployed for political advertising. As a consequence, it should be equally unsurprising that we are witnessing a dramatic increase in AI regulation in the context of politics.
Although the technology is still in its relative infancy, 15 states have already enacted laws governing its use in AI political advertising, and legislation is pending in an additional 16 states and at the federal level. Companies looking to incorporate AI into their political advertising campaigns should therefore be aware of these regulations and what they require.
State AI Laws
As previously stated, 15 states have already regulated the use of AI in political communications or advertisements. Three of those states (Alabama, Mississippi, and Texas) have enacted sweeping bans prohibiting the use of deceptive media (including AI) in political advertising.
The remaining states (Florida, Indiana, Michigan, New York, Oregon, Utah, Wisconsin, Arizona, Colorado, Idaho, Minnesota, New Mexico, and Washington) have passed laws that require political advertisers to include a disclaimer in ads produced using synthetic media, including AI.
The specific content of disclaimers, and when such disclaimers are required, vary from state to state, but a brief review of recently enacted legislation in New Mexico, Florida, Utah, and Michigan helps illustrate how these various disclosure statutes operate.
New Mexico: HB-182, which went into effect on May 1, 2024, amends and enacts sections of the New Mexico Campaign Reporting Act (NMSA 1978 § 1-19-26), by adding AI disclaimer requirements for political advertisements containing materially deceptive media and creates the crime of distributing or entering into an agreement to distribute materially deceptive media. For the purposes of the statute, "Materially deceptive media" means an image, video, or audio that: (i) depicts an individual engaged in conduct or speech in which the depicted individual did not engage; (ii) was published, disseminated, distributed, or displayed to the public without the consent of the depicted individual; and (iii) was produced in whole or in part by using AI.
The required disclaimer is: "This [image/video/audio] has been manipulated or generated by artificial intelligence."
Florida: Florida's CS/HB 919 - Artificial Intelligence Use in Political Advertising went into effect on July 1, 2024, and requires the following disclaimer be included on specified forms of political advertisements created with AI: "Created in whole or in part with the use of generative artificial intelligence (AI)."
Utah: SB-131, which went into effect on May 1, 2024, requires that political advertisements and related communications created using synthetic media (that which is substantially produced using AI) carry one of the following disclaimers depending upon the nature of the communication.
Audio only ads that contain AI-generated voices must include the following audio statement at the beginning and end of the ad: "Contains content generated by AI." Visual ads that contain AI-generated media must display specific disclosures depending upon whether the ad is an image or a video. Image-only ads must include a disclaimer stating: “This image generated by AI.” Disclaimers required for videos depend upon whether the AI-generated content includes images, audio, or both.
Michigan: HB-5141, which went into effect on February 13, 2024, requires that prerecorded telephone messages generated in whole or in part by AI used in support of political campaigns must carry the following disclaimer: "This message was generated in whole or substantially by artificial intelligence." Qualified political advertisements created using AI also must state that they were generated in whole or substantially by AI.
Of the 14 states with pending legislation governing AI-generated political advertisements, those advanced in New Hampshire, Illinois, Ohio, and Pennsylvania seek to impose strict prohibitions on using any type of deceptive media, while the rest propose disclaimer requirements similar to those discussed above.
AI Regulation at the Federal Level
While Congress debates the adoption of statutes governing the use of AI in a variety of contexts, the Federal Communications Commission (FCC) has already started using its existing authority to curb the misuse of AI in political advertising.
As detailed in prior articles, the FCC has already launched two enforcement actions stemming from the misuse of AI-Generated Political Advertisements during the New Hampshire primary election, when thousands of people received calls featuring what appeared to be the voice of President Joe Biden urging them not to vote.
As a result, the FCC issued proposed hefty forfeiture penalties totaling $8 million against the political operative who arranged for the calls and the telecommunications carrier that originated them. In February 8, 2024, the FCC unanimously adopted a Declaratory Ruling that officially held AI-generated voices to be “artificial” as that term is used in the Telephone Consumer Protection Act (TCPA), and thus the legal equivalent of a robocall. The proposed penalties are based on existing FCC authority to deal with robocalls under the Truth in Caller ID Act of 2009 and the Pallone-Thune Telephone Robocall Abuse Criminal Enforcement and Deterrence Act (TRACED Act).
AI-Generated Political Advertisements - Looking Forward
AI regulation is very much a work in progress and grows ever more complex as each new piece of legislation is passed.
Political advertisements that incorporate AI-generated images, voices, or video are already subject to a certain degree of regulation at the state level, and companies that use the technology in connection with political campaigns should take pains to ensure that the final version of any ad includes whatever disclaimers are currently required under state law, and to keep an eye on new requirements as they come into effect.
