The FCC last week released a highly anticipated Notice of Proposed Rulemaking (“NPRM”) seeking comment on proposed disclosure requirements for political ads containing AI-generated content. The item was adopted earlier this month by a 3-2 party-line vote, nearly two months after FCC Chairwoman Rosenworcel first announced its circulation among the commissioners for consideration.
As discussed in more detail below, the proposed rule would require radio and TV broadcasters to (1) inquire of any person making a request to buy airtime for political advertising whether the ad contains AI-generated content; (2) make on-air disclosures of AI use with regard to political ads containing AI-generated content immediately before or during their airing; and (3) include a disclosure of AI use in the station’s Political File records for each such ad buy. While this post focuses on the NPRM’s broadcast-specific proposals, we note that it proposes similar obligations for cable operators, DBS providers, and SDARS licensees engaged in program origination, as well as for Section 325 permit holders (those authorized to export programming for transmission back into the U.S.).
Aware that such rules might conflict with similar efforts by states and other federal agencies, the NPRM characterizes the proposed disclosure requirements as a “complement” to efforts to regulate AI in political advertising that are underway in various states and at the Federal Election Commission (“FEC”), which we wrote about here and here. However, FEC Chairman Sean Cooksey made his contrary views clear in a letter last month to FCC Chairwoman Rosenworcel in which he stated that the FEC has exclusive jurisdiction in this area and “the FCC lacks legal authority to promulgate conflicting disclaimer requirements only for political communications.”
The proposal would require broadcasters to do the following:
Duty of Inquiry: Broadcasters would need to inform each political advertiser, at the time the station agrees to air a political ad, of the requirement that stations must air a disclosure for any ad that includes AI-generated content and then inquire of the buyer as to whether the ad includes such content. While styled as a “simple inquiry,” the NPRM acknowledges various challenges that are likely to arise. It seeks comments on how to deal with such situations, including, for example, where a station is working with a media placement agency that had no role in the creation of the ad and which may not know whether it includes AI-generated content, or the station receives political content from a network or syndicator and has no direct contact with the advertiser.
On-Air Disclosure: A broadcast station that receives a candidate or issue ad containing AI-generated content would need to air a disclosure immediately before or during the ad to inform viewers of the ad’s use of AI. The proposal contemplates and seeks comment on the following standardized language for the disclosure: “[The following]/[this] message contains information generated in whole or in part by artificial intelligence.” Once again, the NPRM acknowledges there are challenges broadcasters will face in complying with the proposed rule. These include (a) what should a station do if it has received no response to its inquiry about AI use; (b) what should a station do if it was told by the person or entity buying the time that an ad contains no AI-generated content and is later informed by a credible third party that the ad does include AI-generated content (and who should qualify as a “credible third party?”); and (c) what should a station do when it receives political programming through a network and lacks any information from the advertiser on AI use as well as the ability to insert a disclosure in network-delivered programming? The NPRM seeks comment on these and many other issues that may affect a station’s ability to comply with the proposed disclosure requirement.
Online Disclosure: Adding yet more to the burden on broadcasters, the NPRM proposes requiring broadcasters to include in their online Political Files the following written disclosure for each political ad containing AI-generated content: “This message contains information generated in whole or in part by artificial intelligence.”
Because of the FCC’s limited jurisdiction, the proposed rules would apply only to certain FCC-regulated entities, doing nothing to address the use of AI in political ads that voters see and hear on social media or elsewhere. As a result, it would impose a significant burden on regulated entities while leaving unregulated entities like social media—the primary source of deceptive political information—completely unregulated. This would incentivize advertisers to put their AI ads on any media other than radio and TV, both because of their desire not to include disclosures and the added bureaucracy/delay involved in the multi-step process stations would need to follow with advertisers to determine if a disclosure is needed (and the added time needed to then insert such a disclosure).
In addition, having disclosures on radio and TV political ads but nowhere else is a sure formula for increasing the harmful impact of AI in online political ads. Seeing such disclosures on radio and TV, the public will think such disclosures are required for all media, causing the public to give undue weight to deceptive social media ads using AI simply because those ads don’t have a disclosure (and therefore must not have any AI content in them).
Finally, there are yet more pragmatic problems with the proposal. The notion of stations selling airtime for 30 or 60-second spots and then surrendering more airtime for a disclosure preceding the spot is unrealistic, both from a financial perspective (stations have only so much airtime for ads and cannot just add more pages/links for ads like a website) and from a scheduling perspective (commercial breaks are of a set duration; you can’t add a five second disclosure without cutting into programming or someone else’s ad).
Perhaps most importantly, the NPRM proposes a requirement that would conflict with those adopted by a number of states while also lacking the universal application that a rule adopted by the FEC or legislation enacted by Congress would have. If the NPRM’s proposed rule is adopted, it will be challenging for broadcasters in general to implement, but for those in states that have already adopted laws addressing the use of AI in political ads, they may find that the FCC rule either conflicts with state law, or requires somewhat different disclosures than the state does. In the latter case, multiple disclosures may leave little room for the political message, with the variety of disclosure messages leaving the public even more thoroughly confused. For example, most state law disclosure requirements apply only to political ads containing deceptive AI, whereas the FCC proposal would apply to any use of AI, even if it is simply used to adjust the contrast in video content.
Broadcasters may find some comfort in the likelihood that the FCC will be unable to complete this rulemaking and require compliance with the new disclosure requirement before the upcoming November election. The NPRM is subject to a 45-day comment and reply cycle—which has not yet opened—and any final rules would be expected to undergo a Paperwork Reduction Act review by the Office of Management and Budget.
Anticipating First Amendment challenges to this proposed regulation of political speech, the NPRM includes a lengthy constitutional discussion seeking to justify the FCC’s authority. The FCC asserts that the proposed rules have been drafted to withstand any level of First Amendment scrutiny—heightened rational basis, intermediate scrutiny, or strict scrutiny—but invites comment on its analyses and its statutory authority to promulgate such rules.
As indicated by FEC Chairman Cooksey’s letter and the dissenting statements of Commissioners Carr and Simington, this proposal will be hotly debated by interested parties spanning various industries and political affiliations. Comments will be due 30 days after publication of the NPRM in the Federal Register, with Reply comments due 15 days later.
[View source.]