Upholding the First Amendment, a senior US district judge has blocked California's deepfakes law, AB 2839, which was designed to stop deceptive AI-generated content from impacting election outcomes.
Last month, a parody video creator known as "Mr Reagan" on YouTube and X (formerly Twitter), Christopher Kohls, sued to block the law. He claimed that AB 2839 was unconstitutionally targeting his protected satirical content—including a fake Kamala Harris video that Elon Musk retweeted, agitating California governor Gavin Newsom.
Kohls urged Judge John Mendez to grant a preliminary injunction, arguing that the law was unconstitutionally vague and allowed anyone who watched his parody videos to sue him, claiming they'd been deceived by his content.
In an order Wednesday, Mendez granted the preliminary injunction, agreeing with Kohls that "the statute infringes on his right to free speech and is unconstitutionally vague." Although the judge acknowledged that the government had a valid interest in protecting election integrity, almost no part of the deepfakes law survives strict constitutional scrutiny, Mendez said, suggesting that the law was not severable and could not be salvaged.
Rather than trying to "enforce silence," California officials would be better off combating deepfakes with "more speech," Mendez said. "Especially as to political speech, counter speech is the tried-and-true buffer and elixir, not speech restriction."
"Victory!" Kohls posted on X. "Lawsuit against Newsom has been won."
Musk predictably joined in celebrating the win, responding, "Congratulations! Score one for the people’s right to free speech."
This acknowledgment clearly pleased Kohls, who responded, "Thank you, sir. We should go to war together again sometime."
A First Amendment scholar and senior fellow with the non-partisan think tank The Future of Free Speech, Jeff Kosseff praised the ruling as setting important precedent regarding AI-generated content online.
"The California law is a clear violation of the First Amendment, so it is not surprising that Judge Mendez quickly blocked it," Kosseff said. "The opinion is a well-reasoned application of First Amendment law and recognizes that government regulation is not an appropriate solution to any challenges that might arise from deepfakes. As lawmakers across the country consider legislation to crack down on political falsehoods, Judge Mendez’s opinion serves as a reminder that such efforts often face insurmountable hurdles, and for good reason."
AI doesn’t abolish right to lampoon government
California's government had tried to argue that the law worked similarly to a defamation law, protecting candidates from deceptive personal attacks during critical election periods. But Mendez noted that AB 2839 "does not use the word 'defamation' and by its own definition extends beyond the legal standard for defamation to include any false or materially deceptive content that is 'reasonably likely' to harm the 'reputation or electoral prospects of a candidate.'"
"At face value, AB 2839 does much more than punish potential defamatory statements since the statute does not require actual harm and sanctions any digitally manipulated content that is 'reasonably likely' to 'harm' the amorphous 'electoral prospects' of a candidate or elected official," Mendez wrote.
Instead, the statute's broad scope dangerously makes government officials arbiters of truth regarding any deepfakes or content that "a reasonable person" might consider deceptive content.
"Even artificially manipulated content that does not implicate reputational harm but could arguably affect a candidate’s electoral prospects is swept under this statute and subject to civil liability," Mendez said. He further criticized the law as "lacking any objective metric" for when content risks undermining confidence in elections or ruining a candidate's election prospects.
"Almost any digitally altered content, when left up to an arbitrary individual on the Internet, could be considered harmful," Mendez said, even something seemingly benign like AI-generated estimates of voter turnouts shared online.
Additionally, the Supreme Court has held that "even deliberate lies (said with 'actual malice') about the government are constitutionally protected" because the right to criticize the government is at the heart of the First Amendment.
"These same principles safeguarding the people’s right to criticize government and government officials apply even in the new technological age when media may be digitally altered: civil penalties for criticisms on the government like those sanctioned by AB 2839 have no place in our system of governance," Mendez said.
According to Mendez, X posts like Kohls' parody videos are the "political cartoons of today" and California's attempt to "bulldoze over the longstanding tradition of critique, parody, and satire protected by the First Amendment" is not justified by even "a well-founded fear of a digitally manipulated media landscape." If officials find deepfakes are harmful to election prospects, there is already recourse through privacy torts, copyright infringement, or defamation laws, Mendez suggested.
Kosseff told Ars that there could be more narrow ways that government officials looking to protect election integrity could regulate deepfakes online. The Supreme Court has suggested that deepfakes spreading disinformation on the mechanics of voting could possibly be regulated, Kosseff said.
Mendez got it "exactly right" by concluding that the best remedy for election-related deepfakes is more speech, Kosseff said. As Mendez described it, a vague law like AB 2839 seemed to only "uphold the State’s attempt to suffocate" speech.
Parody is vital to democratic debate, judge says
The only part of AB 2839 that survives strict scrutiny, Mendez noted, is a section describing audio disclosures in a "clearly spoken manner and in a pitch that can be easily heard by the average listener, at the beginning of the audio, at the end of the audio, and, if the audio is greater than two minutes in length, interspersed within the audio at intervals of not greater than two minutes each."
The audio disclosure is unlike the law's other disclosure requirement that Kohls had complained required text so large that it would block the entire screen for the entire duration of his videos.
Mendez agreed with Kohls that the text disclosure requirement "is not reasonable because it almost certainly 'drowns out' the message a parody or satire video is trying to convey," while the audio disclosure may be considered "compelled speech" but would not necessarily be "unduly burdensome." Kosseff suggested both requirements risked compelling speech that altered the satirical content.
"The court has long held that the First Amendment protects not only against restrictions on speech but against compelled speech," Kosseff told Ars. "That sort of compelled speech really does change the character of something like parody."
But whether the audio disclosure would fly or not, the law falls apart without every other section, Mendez said, having "preliminarily determined that the rest of AB 2839 is still unconstitutional."
Calling AB 2839 "a blunt tool that hinders humorous expression and unconstitutionally stifles the free and unfettered exchange of ideas which is so vital to American democratic debate," Mendez granted the preliminary injunction to stop the state from causing "irreparable injury" to people like Kohls using deepfakes to criticize the government.
In his complaint, Kohls insisted that "making fun of presidential candidates and other public figures is an American pastime," and his deepfakes fall "squarely within this tradition."
Regarding the text disclaimer that threatened to water down satirical content like Kohls' (which Mendez funnily noted Kohls "alleges" to be humorous), Mendez warned that "the obstructiveness of this requirement is concerning because parody and satire have relayed creative and important messages in American politics."
