29 October 2024
A deepfake video circulates on the internet just weeks before a California gubernatorial election, showing a prominent candidate in a closed-door meeting with powerful corporate officials. The candidate is seen and heard accepting large bags of cash in exchange for policy promises. The video looks incredibly realistic, and though no such meeting took place, it contains subtle details—including being set in the candidate’s office and using what sounds like the actual candidate’s voice. The person who made the deepfake claims it is a satirical commentary on corporate influence in politics. However, the damage to the candidate’s reputation is irrevocable, as a significant portion of the electorate believes the footage is real despite the expressed satirical intent.
On September 17, 2024, California’s Defending Democracy from Deepfake Deception Act of 2024 passed both California’s Senate and Assembly and was officially signed into California law by Gavin Newsom [1]. The bill prohibited the creation and distribution of “materially deceptive” election-related content, specifically deepfake and AI-generated media, within 60 days of an upcoming California election. Planned to start in this year’s election cycle, the bill mandated that large online and social media platforms take measures to prevent the uploading of deepfakes that qualify as “materially deceptive” content, particularly when such content targets a specific candidate by depicting false information that could damage their reputation.
However, the bill excluded certain content from being flagged as “materially deceptive content.” This exemption includes satire. The exact language from the bill states: “This chapter does not apply to any of the following: (c) materially deceptive content that constitutes satire or parody.”
In July 2024, Elon Musk reposted an AI-edited video of Kamala Harris titled “Kamala Harris Campaign Ad PARODY” that mimicked a real campaign ad, using AI-generated audio of the Vice President’s voice to spread questionable claims about her [4]. In response to the popularity of the video on social media, Gavin Newsom pushed for a Deepfake Bill in hopes of restricting similar content that he believed would deceive the public [2].
Christopher Kohls, the original creator of the video, and Elon Musk, who reposted it on X, both argued that any restrictions on the video would infringe on their free speech rights, claiming it is parody and satire. On September 18, 2024, Kohls filed a lawsuit, alleging that Governor Newsom’s actions are unlawfully restricting his right to political satire and parody, which is protected under the First Amendment.
The key issue lies in how “satire” and “parody” are defined under the Deepfake Bill. While the bill exempts satirical and parody content, Governor Newsom’s restriction of the AI-altered video of Kamala Harris suggests that he views the video as falling outside those protected categories. On the other hand, Christopher Kohls, the video’s creator, and Elon Musk claim the video is exactly the type of content that the exemption has granted. The central question is now whether the video genuinely fits these categories or crosses the line into “materially deceptive content.”
On October 2, a verdict was reached in Christopher Kohls’s lawsuit against the California law, with Senior U.S. District Judge John A. Mendez ruling in Kohls’ favor. Mendez indicated that the Kamala Harris video would fall under satire. He also ruled that the law, as currently written, is overly broad and could result in significant overreach by state authorities. As of October 4th, the bill is currently blocked by a preliminary injunction. While this means the bill could potentially be reinstated, the judge’s comments suggest that revisions may be necessary before it can move forward.
While California’s new bill would have set an important precedent for deepfake regulation, it fell short of adequately defining what constitutes satirical content. As realistically generative technology becomes more accessible to the general public, clearer parameters are needed to guide content creators as to whether they are or are not breaking the law with their social media posts.
In the 1988 Supreme Court case Hustler v. Falwell, the Court ruled that a parody, which no reasonable person would take seriously, is protected under free speech. The case involved Hustler Magazine publishing an article about political figure Jerry Falwell, in which it made outlandish claims, such as him having a drunken, incestuous encounter with his mother in an outhouse. The court found that, despite the deceptive nature of the content, it qualified as satire because it was so absurd that no reasonable person would believe it, and there was no malicious intent.
In any form, whether it be a newspaper illustration, animated cartoon, or online video, satirical content assumes a modicum of intelligence from the viewer to recognize its nuance and underlying commentary. However, when deepfakes and AI-altered images use someone’s likeness to create portrayals of that person that are indistinguishable from reality, how do we draw the line between satire and maliciously deceptive content? How do we more clearly, in the eyes of the law, classify between satire and deceptive manipulation when such convincing tools are available to content creators?
The key distinction between earlier landmark satire cases, such as Hustler Magazine v. Falwell, and contemporary cases involving social media lies in the established reputation of the content creator. In Hustler Magazine, the publication was widely recognized as a provocateur, and its readers knew to expect absurd, humorous content that was not intended as serious news reporting. With social media, however, newer creators often lack such established reputations or large fanbases, meaning that viewers have no preconceived expectations about their content. As a result, the ability to recognize satire now depends more on the content itself rather than the creator’s reputation, which presents a challenge for discerning satire in isolated instances.
One way to mitigate potential harm to a candidate is to include a provision in the California bill that requires any expressed satirical content that uses deepfakes or AI-generative technology to display a clear, easily visible watermark or disclosure stating the following: “Audio or video representations in this post are intended for satirical purposes only.”
The watermark and disclosure should be prominently displayed within the video itself rather than in the title or comments. This ensures that, even if the video is reposted, such as in the case of Elon Musk’s reposting of the Kamala Harris video, or uploaded to other platforms, the watermark will remain visible and attached to the video regardless of where it is shared.
While a watermark can mitigate some confusion, it does not fully address the challenge of defining satire, especially if the uploader labels it as such. For instance, even with a watermark, Gavin Newsom might still perceive the AI-generated parody video of Kamala Harris as deceptive and harmful. This highlights the need for a new, clear criterion for identifying satire—one that provides a rigorous scrutiny test and can be applied consistently by the Supreme Court and other judicial bodies. As deepfake technology advances, we can anticipate an increase in cases involving such content. This makes it crucial to establish a robust and indisputable standard for distinguishing satire from potentially misleading material.
As it currently stands, the California Deepfake Bill is ambiguous and ineffective regarding who and what constitutes the definition of satire. Is it up to the state to define satire, or can creators declare their own content as satirical? The bill’s provisions leave unclear whether satire remains protected if it involves replicating a candidate’s voice or placing the candidate in a fictional photorealistic situation. Can a satirical, deepfake video include political endorsements? Can it make use of the candidate’s family or staff members’ likenesses? Can it include niche humor that would only be found funny by a specific group? Until California’s Defending Democracy from Deepfake Deception Act of 2024 better defines satire, these questions will remain up in the air and many more lawsuits are surely to be expected.
If California were to revise the satire exemption of the bill, it could be amended as follows:
“In determining whether content constitutes satire, the following criteria shall be considered: [1] whether the level of visual or auditory realism does not outweigh, overpower, or override the absurdity of its narrative; [2] whether the satirical intent is evident through the narrative or objectively implicit in its messaging; and [3] whether the satirical element persists independently of the hyperrealistic medium—i.e., whether the satirical discernment remains clear when the narrative or script is separated from its realistic presentation. The narrative should always remain the primary focus and must not be dominated to the point that it becomes secondary or overshadowed by the viewer’s perception of realism.”
Image via Pexels Free Photos
Works Cited
[1] “AB 2655: Defending Democracy from Deepfake Deception Act of 2024.” Digital Democracy. February 12, 2024. Accessed August 27, 2024 https://digitaldemocracy.calmatters.org/bills/ca_202320240ab2655
[2] White, Jeremy “Gavin Newsom signs election ‘deepfake’ ban in rebuke to Elon Musk.” Politico. September 17, 2024. Accessed September 18, 2024. https://www.politico.com/news/2024/09/17/newsom-signs-election-deepfake-ban-00179557
[3] Korte, Laura “Creator of Kamala Harris parody video sues California over election ‘deepfake’ ban.” Politico. September 18, 2024. Accessed September 18, 2024. https://www.politico.com/news/2024/09/18/california-deepfake-ban-lawsuit-harris-00179975
[4] Kohls, Christopher (@MrReaganUSA) “Kamala Harris Campaign Ad PARODY”-X. July 26, 2024. Accessed August 27, 2024. https://x.com/MrReaganUSA/status/1816826660089733492
Policy & Political Review 