A federal judge blocked one of California's new AI laws on Wednesday, less than two weeks after it was signed by Gov. Gavin Newsom. Shortly after signing AB 2839, Newsom suggested that the signature could be used to force the removal of an AI deepfake of Vice President Kamala Harris that was reposted by Elon Musk. A petty online fight broke out between them). But a California judge just ruled that the state can't force people to delete election deepfakes, at least not yet.
AB 2839 targets distributors of AI deepfakes on social media. This is especially true if the post resembles a political candidate and the poster knows it's a fake that could confuse voters. This law is unique in that it goes after those who spread AI deepfakes, rather than the platforms on which they appear. AB 2839 would give California judges the power to order posters of AI deepfakes to take them down or face potential fines.
Perhaps unsurprisingly, the original poster of that AI deepfake, an X user named Christopher Coles, filed a lawsuit calling California's new law unconstitutional just one day after it was signed into law. Coles' lawyers wrote in the complaint that the Kamala Harris deepfake is satire that should be protected by the First Amendment.
On Wednesday, U.S. District Judge John Mendez sided with Coles. Mr. Mendez issued a preliminary injunction temporarily blocking the California attorney general from enforcing the new law against Mr. Coles and others, except for audio messages covered by AB 2839.
Read for yourself what Judge Mendez said in his ruling.
“Almost any digitally altered content could be deemed harmful if left in the hands of any individual on the internet. For example, approximate AI-generated numbers on voter turnout could be On the other hand, many “harmful” depictions, if shown to a variety of individuals, can ultimately be viewed as false content that reasonably undermines confidence in the election results based on the election results. It may not affect the election or undermine confidence in the election at all. As the plaintiffs persuasively point out, AB 2839 “relies on a variety of subjective terminology and clumsily worded mens rea,” thereby eliminating vast amounts of political and constitutionally protected speech. It has the effect of suggesting…
[W]Well-founded fears of a digitally manipulated media environment may be justified, but this fear has led legislators to limit the long tradition of criticism, parody, and satire protected by the First Amendment. This does not give permission to interfere with. YouTube videos, Facebook posts, and Protected. Other statutory causes of action, such as privacy torts, copyright infringement, and defamation, may be brought against the public whose reputation may have been harmed by artificially altered depictions spread by satirists and opportunists on the Internet. We have already filed lawsuits against people and individuals…
This record shows California has a strong interest in preserving election integrity and addressing manipulated content. But California's concerns and the challenges the state faces are at stake because of the gravity of the First Amendment values at issue and how plaintiffs and other similarly situated content creators have had their speeches chilled out. This is minimal compared to the ongoing constitutional violations we are experiencing. ”
In essence, he judged that the law was too broad in its terms and could potentially allow national authorities to take serious steps in determining what speech is allowed.
This is a preliminary injunction, so we'll have to wait and see whether the California law will actually be permanently blocked, but either way it's unlikely to have much of an impact on next month's election. AB 2839 is one of 18 new AI-related laws that Newsom signed last month.
Nevertheless, this is a big victory for Elon Musk's camp, which puts up free speech posters about X. Days after Newsom signed AB 2839 into law, Musk and his usual allies posted a series of AI deepfakes testing California's new law.