Artificially Intelligent or Artificially Malevolent?: The Free Speech Cost of Taming AI

Artificial intelligence, also known as AI, has experienced an unprecedented surge in its scope of knowledge and power. In recent years, it has brought excitement to the fields of computer science and information systems as well as our own homes in the form of personal assistants like Alexa and Siri [1]. Nonetheless, AI has sparked controversy, from its aid in academic dishonesty to more serious issues such as identity theft and libel in the form of “deepfakes”. The phenomenon of “deepfakes” refers to media such as videos, pictures, or voice recordings generated by artificial intelligence to imitate or depict real or non-existent people [2]. Deepfakes have gained attention for their use in creating celebrity pornographic photos, bullying, financial fraud, identity theft, and fake news. Intellectuals and academia have raised questions about the implications deepfakes could have on misinformation, hate speech, and interference with elections. In response to these inquiries, the information technology industry along with other government leaders and agencies have formulated recommendations and actions to detect and combat their use. 

Gavin Newsom, the Governor of California, has made headlines as one of the first government leaders to address the issue of AI. He recently signed into law two significant pieces of legislation aimed at criminalizing deep-fake parodies. The first is the "Defending Democracy from Deepfake Deception Act of 2024" (formally known as AB 2655) which “seeks to regulate and ban “disinformation powered by generative AI because of ‘California’s compelling interest in protecting its free and fair elections” [3]. This act requires larger social media companies such as X and YouTube to censor “AI-generated parody videos of candidates and election videos that could ‘harm the reputation or electoral prospects of a candidate’ or ‘falsely undermine confidence in the outcome of one or more election contests’ [3].  In addition to this policy, Newsom passed the "Elections: Deceptive Media in Advertisements Act of 2024” (formally known as AB 2839) which prefaces by asserting the dangers of generative AI as it advances the promotion of misinformation. The act’s primary purpose is to “provide consumers with factual information about the inauthenticity of particular images, audio, video, or text content to prevent consumer deception” [3]. The main stakeholder in this bill was purported to be advertising. However, the bill extends into the context of all election communication which includes all speech related to or concerning a candidate running for office. This speech extends beyond television advertisements or billboards and also concerns speech across internet platforms and all forms or modes of social media. 

Both pieces of legislation have sparked discussions about their potential impact on free speech. Thus, Newsom's proactive stance has not been without controversy, as the bills have already faced legal challenges. Christopher Kohls, a conservative social and political commentator who makes content under the alias "Mr. Reagan",  has taken legal action against the state of California by filing a complaint through the attorney general. While the lawsuit targets California as a whole, it represents an indirect challenge to Governor Newsom's policies. The case Christopher Kohls v. Rob Bonta, in his official capacity as Attorney General of the State of California (2024), has prompted the courts to consider if these two bills Newson signed into law were a breach of First Amendment Rights to free speech. 

On July 26th, 2024, Elon Musk posted a parody of Kamala Harris's campaign video on his platform X (formerly known as Twitter). The video, created by Kohls, mocked Harris' mannerisms and common sayings using AI-generated voice recordings that closely resembled her voice. In the video, "Harris" referred to herself as a "diversity hire" and made other controversial statements that would never be included in a presidential campaign advertisement. Two days after the video's release, Gavin Newsom retweeted, "Manipulating a voice in an 'ad' should be illegal. I'll be signing a bill in a matter of weeks to make sure it is." Two months later, on September 17th, 2024, Newsom signed into law two bills aimed at addressing this issue. In retaliation, Kohls has filed a complaint for declaratory and injunction relief with the district court of Eastern District of California Sacramento Division. Kohls argues that Political satire is a fundamental First Amendment right and he filed the claim to defend all Americans’ right to satirize politicians. 

Kohls is referring to the landmark Supreme Court case of Hustler v. Falwell, 485 U.S. 46 (1988). In this case, Hustler Magazine ran a full-page “parody” advertisement against Jerry Falwell Sr. – an American Baptist pastor, televangelist, and conservative activist – depicting him as an incestuous drunkard who had intercourse with his mother in an outhouse. In response, Falwell sued Hustler magazine and its publisher Larry Flint for damages on the account of infliction of emotional distress, libel, and invasion of privacy. In the preliminary circuit court trial, the jury found Hustler liable on the emotional distress claim and Falwell was awarded a total of $150,000 in damages [5]. Hustler appealed this ruling to the Supreme Court which ruled in an 8-0 decision in favor of the magazine. The court held that the emotional distress inflicted on Falwell due to the parody advertisement was not an adequate reason to restrict the First Amendment right to speech that is offensive to public figures and officials. The interest in protecting free speech is more important than the “state's interest in protecting public figures from patently offensive speech, so long as such speech could not reasonably be construed to state actual facts about its subject” [5].

This ruling is critical to the case Kohls has filed against the State of California because his video treads on the same lines of criticism as the Hustler magazine advertisement. The video he posted on YouTube portraying Kamala Harris is explicitly labeled as a “parody” and acknowledges that “sound or visuals were significantly edited or digitally generated”. However, the problem lies in the fact that the speech in some regards could be construed to state facts about Harris, her beliefs, and her campaign platform. The video’s voiceover closely resembles Harris’ voice and the production itself almost identically mirrors the theme and aesthetic of Harris’ original campaign ads, often using clips from her actual campaign videos. Though the comedic effect of the video becomes clear once Harris identifies herself as a “diversity hire”, the video if not watched in its entirety is brimming with false claims [3]. In modern times where misinformation is spreading like wildfire and society is very polarized these ads if not explicitly stated to be “parodies” can be dangerous. These implications beg the question of whether it is constitutional to bar the production of deepfakes concerning political candidates. 

The difference between Christopher Kohls v. Rob Bonta, in his official capacity as Attorney General of the State of California and the Hustler Magazine, Inc. v. Falwell, 485 U.S. 46 (1988) case is the legal question. With Hustler, the case beckons the question of whether or not the First Amendment's freedom of speech protection extends to the creation of satirical (but offensive) statements about public figures, resulting in emotional distress, discomfort, and invasion of privacy. Whereas, in this complaint filed against the state of California the matter subject to review is whether the First Amendment's protection of freedom of speech includes the dissemination of misinformation and defamation of political candidates.

The body of the AB 2655 bill – also known as the "Defending Democracy from Deepfake Deception Act of 2024” — states that one of the bill’s purposes is to prevent false media that could “harm the reputation or electoral prospects of a candidate” which is questionable [3]. This reasoning draws connections to the principles of the ideas of libel – a published false statement that is damaging to a person's reputation– and the landmark decision of the Supreme Court in New York Times v Sullivan 376 U.S. 254 (1964). In this decision, the court established what has come to be known as the actual malice rule. This principle refers to the means that public officials or figures who are suing for libel must prove that the statements made by a speaker were done with bad intentions. When a statement concerns a public figure, “the Court held, it is not enough to show that it is false for the press to be liable for libel but instead, the target of the statement must show that it was made with knowledge of or reckless disregard for its falsity” [6]. By these standards Kohls' video – although not completely free from negative intent – is worthy of being free from censorship. This is because he did clarify that the footage aimed towards Harris’ platform was false and for parody purposes. Furthermore, given the fact that Harris (and other political candidates) is a known public figure, the level of fault for defamation is much higher than the average civilian. Due to Harris being a public figure, she does have greater access to the media to counter defamatory or false statements (such as the one Kohls published) and, to a certain extent, seek out public acclaim and assume the risks of having a high profile. So, although Newsom’s policies are well-intentioned it lies in a legal grey area because it unsoundly inflicts on the rights of American citizens to criticize and satirize politicians. 

However, it can be argued Kohls was acting maliciously and did intend to “cause harm” to the image of Harris with his particular statement of her being a “DEI hire”. That statement is charged with racist undertones and intentionally plays into the long-standing discriminatory narratives that aim to discredit people of color in positions of power. Many claims made in the video raise complex questions about where protected political satire ends and malicious racial and gender-based harassment begins. Kohl's content, while packaged as political commentary, can serve to amplify harmful stereotypes and undermine the credibility of public officials based on their race and gender rather than policy or actions – further contributing to the spread of misinformation. 

Furthermore, Kohls' video is extremely vulnerable to misuse especially given the fragmented state of the current social media landscape. While Kohls may have designated the content in the video as “parody” in its YouTube title, this context fades when the content travels across different platforms (as it did to Elon Musk’s X account). Short clips from his video can be extracted and shared on platforms like TikTok, X, and Instagram while lacking the original disclaimer, transforming an “innocent parody” into damaging material for misinformation. The cross-platform proliferation presents a harsh truth: though the video initially shared may meet the regulations for parody on one platform there is nothing keeping it from upholding the appropriate regulations and disclaimers on another. In this regard I find that the video itself may be worthy of censorship but in a limited capacity. 

The implementation of warning and indication systems by platforms like YouTube represents a potential middle ground between outright censorship and the unregulated spread of misleading content. Through a new likeness management technology YouTube has championed a system “that will enable people from a variety of industries—from creators and actors to musicians and athletes—to detect and manage AI-generated content showing their faces” on the platform [8]. This approach acknowledges both the right to critical and satirical political speech as well as the need to protect against deliberate public manipulation and misinformation.  

However, the effectiveness of such measures to ensure societal preservation remains uncertain. The fast-paced development of AI has brought this topic of free speech within the realm of multimedia into uncharted territory as in previous times satire consisted simply of newspaper articles and less technological media. Deepfakes bring this idea of satire to a new level of sophistication, the line between protected political speech and harmful impersonation grows increasingly blurred. The case Christopher Kohls v. Rob Bonta, in his official capacity as Attorney General of the State of California, is still ongoing and will likely set important precedents for how we balance First Amendment protections with the need to prevent technology-enabled fake news. As AI continues to evolve, the courts and legislators will need to carefully weigh these competing interests while ensuring that protections against misinformation don’t unduly restrict legitimate political discourse. 

 

 

 

[1]‌  The History of AI: A Timeline of Artificial Intelligence.” Coursera, 

[2]  Gao-20-379SP, Science & Tech Spotlight: Deepfakes

[3] Christopher Kohls v. ROB BONTA, in his official capacity as Attorney General of the State of California, and SHIRLEY N. WEBER, in her official capacity as California Secretary of State, 

[4] “Artificial Intelligence, Free Speech, and the First Amendment | the Foundation for Individual Rights and Expression.” 

[5] Hustler Magazine, Inc. v. Falwell, 485 U.S. 46 (1988)

[6] New York Times v Sullivan 376 U.S. 254 (1964)

‌[7] Press, Associated. “San Francisco Files First-of-Its-Kind Lawsuit to Tackle AI Deepfake Nudes - POLITICO.” POLITICO

[8] Hanif, Amjad. “New Tools to Protect Creators and Artists.” Blog.youtube,

 

 

Works Cited

“Artificial Intelligence, Free Speech, and the First Amendment | the Foundation for Individual 

Rights and Expression,” n.d, online at www.thefire.org (visited November, 2024). 

Christopher Kohls v. ROB BONTA, in his official capacity as Attorney General of the State of 

California, and SHIRLEY N. WEBER, in her official capacity as California Secretary of 

State, Complaint For Declaratory and Injunction Relief “DocumentCloud.” Documentcloud.org (visited November, 2024). 

Gao-20-379SP, Science & Tech Spotlight: Deepfakes, www.gao.gov/assets/gao-20-379sp.pdf.

Hanif, Amjad. “New Tools to Protect Creators and Artists.” Blog.youtube, YouTube Official 

Blog, 5 Sept. 2024, online at blog.youtube/news-and-events/responsible-ai-tools/ (visited November, 2024). 

Hustler Magazine, Inc. v. Falwell, 485 U.S. 46 (1988), n.d, Justia US Supreme Court Center, 

online at https://supreme.justia.com/cases/federal/us/485/46/ (visited November, 2024).

New York Times v Sullivan, 376 U.S. 254 (1964), n.d, Justia US Supreme Court Center, 

online at https://supreme.justia.com/cases/federal/us/376/254/ (visited November, 2024).

 ‌Press, Associated. “San Francisco Files First-of-Its-Kind Lawsuit to Tackle AI Deepfake Nudes 

- POLITICO.” POLITICO, Politico, 17 Aug. 2024

The History of AI: A Timeline of Artificial Intelligence.” Coursera, online at 

www.coursera.org/articles/history-of-ai (visited November, 2024).