Elon Musk’s Reposts of Kamala Harris Deepfakes Could Face Scrutiny Under New California Law

California's new law introduces significant legal consequences for social media users who post or share AI-generated deepfakes aimed at misleading voters about upcoming elections. Governor Gavin Newsom announced that Assembly Bill 2839 (AB 2839), which took effect immediately upon his signature on Tuesday, is designed to address the dissemination of deceptive content, even targeting influential figures like Elon Musk.

"I just made it illegal to knowingly distribute deceptive election materials in California," Newsom tweeted, referencing a particular AI deepfake Musk shared earlier this year that falsely suggested Vice President Kamala Harris referred to herself as an incompetent candidate and a diversity hire.

"You cannot knowingly share any advertisement or election communication that includes materially false information — such as deepfakes," Newsom emphasized in his tweet.

The new legislation specifically targets the sharers of AI deepfakes that resemble candidates on California ballots, holding them accountable if they knowingly share misleading content that could confuse voters. Unlike other laws that may penalize creators or platforms, AB 2839 focuses on individuals who maliciously circulate these harmful fakes. Under this law, anyone who encounters an AI deepfake on social media can seek injunctive relief, allowing a judge to compel the user to remove the content or impose financial damages.

This is one of the most robust laws in the U.S. against the misuse of AI deepfakes in relation to elections, especially critical as the 2024 presidential race approaches.

The California Initiative for Technology and Democracy (CITED), a key sponsor in drafting AB 2839, pointed out that this law applies to all social media users who share election-related AI deepfakes with malicious intentions. "Malice" refers to the poster's awareness that the content is false and intended to confuse voters.

"AB 2839 targets both the creators and distributors of misleading content if it meets the bill’s criteria," Leora Gershenzon, CITED’s policy director, explained in an interview. "This pertains specifically to deceptive content shared with reckless disregard for the truth that could influence electoral outcomes."

When asked about Musk's potential legal liabilities for sharing deepfakes, Newsom did not discount the possibility of consequences. "I think Mr. Musk may not have grasped the seriousness of the situation," Newsom stated at a press conference. "While parody remains valid in California, deepfakes that manipulate electoral outcomes damage our democracy."

This new law explicitly bans election-related AI deepfakes across various media channels, including television, radio, phone communications, texts, and any online platforms. It extends beyond political campaign advertisements, affecting casual social media posts from all users. AB 2839 establishes a critical 120-day window before elections and a 60-day period afterward, imposing stricter guidelines on political discourse involving candidates on social media.

"Our primary objective isn't to penalize but to prevent the distribution of disinformation altogether," Gershenzon noted. "The best outcome would be to deter the creation and sharing of deepfakes that could mistakenly influence our elections."

This law applies to candidates in California's state, local, and federal elections, including prominent figures like Kamala Harris and Donald Trump. Notably, if an AI deepfake includes a clear disclaimer indicating it has been digitally altered, it is exempt from AB 2839's regulations.

In a challenging test of the law, Musk shared the deepfake of Harris mentioned by Newsom, garnering over 31 million impressions on X. He also posted another deepfake of Governor Newsom that received over 7 million impressions the following day.

Musk and X are already navigating other legal challenges related to content moderation. Recently, a Brazilian Supreme Court judge fined the X Corporation for violating that country's ban on the platform. The judge stated that X's inaction against fake news and hate speech risks undermining democracy in Brazil.