California has moved quickly to put new limits on election-related AI deepfakes, and the law is aimed not only at campaign operations or content creators. Under AB 2839, people who knowingly spread deceptive election communications can face legal consequences, including when the content is reposted on social media.
Governor Gavin Newsom has already suggested the law could apply to high-profile reposts by Elon Musk, including AI-generated material resembling Kamala Harris and Newsom himself. The immediate message from California is direct: sharing a fake can matter legally when it is designed to confuse voters.
What AB 2839 changes
AB 2839 went into effect immediately after Newsom signed it on Tuesday. The law targets AI deepfakes connected to elections when the material resembles a candidate on California ballots and the person distributing it knows it is fake and likely to cause confusion.
The law is notable because it focuses on distribution. It can reach people who post or repost deceptive election content, rather than only the people who originally created the AI deepfake or the platform where it appears.
Newsom described the change in a tweet after referencing an AI deepfake that Musk reposted earlier this year. That video made it appear Kamala Harris called herself an incompetent candidate and a diversity hire, which she did not.
"You can no longer knowingly distribute an ad or other election communications that contain materially deceptive content — including deepfakes," Newsom said later in the tweet.
The practical effect is that everyday social media users can be part of the law’s reach. If a person knowingly spreads a fake election communication with the required level of intent, AB 2839 gives others a way to seek court intervention.
Who can act, and what they can seek
Anyone who sees an AI deepfake on social media can now file for injunctive relief. That means a judge could order the person who posted it to remove the content. The law also allows monetary damages against the person who posted it.
California Initiative for Technology and Democracy, known as CITED, helped draft AB 2839. The group told TechCrunch that the law can affect any social media user who posts or reposts election-related AI deepfakes with malice. In the source article, malice is described as meaning the poster knew the content was false and would confuse voters.
CITED policy director Leora Gershenzon explained the standard this way in an interview with TechCrunch:
"[AB 2839] goes after the creators or distributors of content, if the content falls within the terms of the bill," said CITED’s policy director, Leora Gershenzon, in an interview with TechCrunch. "This is materially deceptive content that is distributed knowing it’s false, with reckless disregard of the truth, and is likely to influence the election."
Gershenzon also said the main goal is prevention, not punishment. In her view, the best outcome would be that deepfakes do not fraudulently affect elections in the first place.
Where the law applies
AB 2839 covers election-related AI deepfakes distributed through several channels. The source article identifies TV, radio, phone, texts and any communication "distributed through the internet."
That broad phrasing matters because the law is not limited to political campaign ads. Other laws have focused on campaign advertising, but AB 2839 also reaches posts from ordinary people during a defined election window.
The stricter period runs 120 days before a California election and 60 days after. During that window, the law places tighter limits on what people can post about political candidates when the content is materially deceptive and distributed with the required knowledge or disregard for the truth.
The law pertains to candidates for state and local elections in California. It also covers federal candidates who will appear on California’s ballot, including Kamala Harris and Donald Trump.
There is an important exception. If an AI deepfake includes an obvious disclaimer saying it has been digitally altered, AB 2839 does not apply to it.
Why Musk became an early test case
Musk’s reposts have become a central example because Newsom referenced one of them when discussing the new law. When asked whether Musk could face legal action for reposting deepfakes, Newsom did not rule it out.
"I think Mr. Musk has missed the punchline," said Governor Newsom at a press conference Thursday. "Parody is still alive and well in California, but deepfakes and manipulations of elections — that hurts democracy."
After the law was signed, Musk reposted the deepfake resembling Kamala Harris that Newsom had referenced. According to the source article, that repost amassed more than 31 million impressions on X. Musk also reposted an AI deepfake resembling Governor Newsom on Wednesday, which received more than 7 million impressions.
The situation shows how AB 2839 is likely to be watched in practice. The legal question is not simply whether synthetic political content exists, but whether a person knowingly distributed materially deceptive election content in a way likely to influence voters.
The issue also sits alongside other moderation-related legal pressure facing Musk and X. The source article notes that a Brazilian Supreme Court judge fined the X Corporation on Thursday for skirting the country’s ban on the platform. The judge previously said X’s failure to combat fake news and hate speech is harming Brazil’s democracy.
For California, AB 2839 makes the risk clearer for social media users heading into the 2024 presidential election. Reposting an AI deepfake is no longer just a platform moderation issue when the content falls within the law’s terms. It can become a court matter.