Exposing the Truth Behind Deepfakes in Political Ads

Political advertisements showing clips of newscasts or politicians making speeches and talking to voters are nothing new – and with a little over a year left until the 2024 elections, voters are gearing up to be bombarded with these ads. But the rise of generative artificial intelligence (AI) has raised a new concern about political advertisements: How do we know what’s being shown in these ads is real?

Over the last year, generative AI platforms have quickly become a popular way to create music, artwork, text and videos. At this point, most people have played around with ChatGPT or listened to AI-generated music. Now, political advertisers are joining the generative AI movement and using it to create photos, videos and audio clips that align with their political message. For example, a Ron DeSantis-supporting political action committee (PAC) used generative AI to create a “deepfake” of Donald Trump’s voice attacking the Republican governor of Iowa. Meanwhile, the Republican National Committee used AI-generated images of boarded up storefronts and military on the streets of U.S. cities to show what they envision happening if President Biden is re-elected.

These developments have raised concerns for online platforms, lawmakers and regulators trying to combat misinformation in political advertising.

Google’s New Policy

Google recently announced a new policy aimed at educating viewers when a political advertisement contains AI-generated content. Starting in November 2023, Google will require any political advertisements that feature “synthetic content” (such as AI-generated photos or videos) that “inauthentically represents real or realistic-looking people or events” to include a clear and conspicuous disclosure noting for viewers that it contains AI-generated content.

As an example, Google specified that any ad with synthetic content that “makes it appear as if a person is saying or doing something they didn’t say or do” or that “alters footage of a real event or generates a realistic portrayal of an event to depict scenes that did not actually take place” would need a disclosure.

Notably, Google also clarified that ads that use AI in an “inconsequential” way (such as image re-sizing, color correction, etc.) would not need a disclosure.

Lawmakers and the FEC Look to Regulate AI-Generated Political Ads

Earlier this year, Rep. Yvette Clark introduced the REAL Political Ads Act, which would require a disclaimer on any political ads that use images or video generated by AI, no matter the medium or platform on which those ads appear. Similar legislation was also introduced in the Senate by senators Amy Klobuchar, Cory Booker and Michael Bennett. However, given partisan gridlock in both the Senate and House of Representatives, the future of the Act is unclear.

The Federal Election Commission (FEC) is also taking steps to regulate AI-generated political advertisements. The FEC opened public comment on a petition for the FEC to regulate this content by amending regulations that prohibit a candidate or their agent from “fraudulently misrepresenting other candidates or political parties” to make clear that this prohibition would apply to deliberately deceptive AI-generated campaign ads. Transparency advocates see this as a sign that the FEC is taking the issue seriously.

Other Online Political Ad Regulations

As online platforms, political advertisers and their agencies continue to watch for more movement in the call to regulate AI use in political advertising, these parties should also be aware of the complicated patchwork of laws governing online political advertising more generally.

FEC regulations have long required that political advertisements on television and radio disclose who paid for such ads. Last year, the FEC approved new regulations that expand its disclosure requirements to cover “internet public communications.” This includes “any public communication over the internet that is placed for a fee on another person’s Web site, digital device, application, or advertising platform.” While the exact form and content of the required disclaimer varies depending on the entity that authorizes and finances the advertisement (for example, a PAC, independent expenditure or candidate/campaign), generally speaking, these internet communications will now require a disclaimer stating who paid for and authorized the advertisement. This move closes a loophole that previously exempted these ads from the FEC disclosure requirements, as earlier FEC regulations only required disclaimers on ads “placed for a fee on another person’s Web site.”

Many states have also enacted legislation over the last few years aimed at regulating online political advertisements. These laws vary greatly from state to state, with some simply extending existing requirements for TV and radio advertising to online political advertising. Meanwhile, some go further and impose new requirements (most commonly robust recordkeeping requirements) on online platforms and ad networks.

Because state laws vary so greatly, political advertisers and online platforms should carefully review each advertisement they place or accept to ensure compliance with all applicable legal requirements.

The Bottom Line

• Generative AI tools have made it easier and cheaper for political groups and campaigns to create convincing but fictitious attack ads targeting political rivals.
• There are an array of proposed tactics that would require political advertisers to clearly and conspicuously disclose when generative AI tools were used to create the content in their advertising.
• Publishers and agencies working with political campaigns and operatives to create and distribute political advertising should familiarize themselves with existing rules and requirements and closely watch the rapidly changing developments heading into the 2024 election cycle.