Blog Post
X is suing to stop our anti-disinfo law. Here’s what you should know.
This is all about people power vs. billionaires who want to profit off of lies.
Article
CITED: First Amendment Notes
AB 2839 and AB 2655 are narrowly tailored and carefully crafted to address the worst election disinformation without compromising free speech, inhibiting our political conversation, or inadvertently compromising harmless content.
California Initiative for Technology and Democracy (CITED) First Amendment Notes
Governor Newsom recently signed two pieces of legislation addressing the threat that AI-powered disinformation poses to our elections.
AB 2839 and AB 2655 are narrowly tailored and carefully crafted to address the worst election disinformation without compromising free speech, inhibiting our political conversation, or inadvertently compromising harmless content.
AB 2839
The Limited Nature of AB 2839
AB 2839, effective immediately, prohibits political deepfakes that portray a candidate, an elected official, or elections official as doing or saying something that they did not do or say, or specified election equipment and voting sites in a materially false way. It does this in a way that targets just the worst of election disinformation.
- It applies only to “materially deceptive” deepfake content, meaning it only applies to content that would deceive a reasonable person. Things that are obviously fake, jokey, cartoonish, or satirical are not touched by the bill. (§ 20012(b)(1))
- It requires “malice” on the part of the person distributing the materially deceptive deepfake content, meaning it only applies if the person knows the content is false or acts with reckless disregard as to the truth. (§ 20012(b)(1))
- Parody or satirical political deepfakes that are so hyper-realistic they would fool a reasonable person into believing they are true must inform the viewer that the content was digitally created or manipulated. (§ 20012(b)(3))
- It only applies within 120 days before an election and, for those deepfakes regarding election officials or voting systems, within 60 days after the election. It applies only when the risk of election disinformation is highest, not at all times. (§ 20012(c))
Only Bad Actors Have Anything to Worry About
In short, AB 2839 only prohibits political deepfakes that are intended to deceive and are so sophisticated that they WOULD deceive a reasonable person. (A reasonable person standard is common in the law and courts have no difficulty applying it in various areas.) Unless a social media user or political actor is attempting to lie to and deceive their fellow Americans and fraudulently undermine an election, AB 2839 does not affect them.
What Would AB 2839 Apply to, In Real Terms?
An example of a political deepfake the bill would apply to is the “Joe Biden” robocall that went to New Hampshire Democrats before that state’s primary election, urging them NOT to vote. The fake robocall was intended to deceive voters and sophisticated enough that it would deceive almost anyone.
What are the Remedies in the Law?
AB 2839 allows a recipient or victim of a deepfake that is in violation of the bill to use an expedited process to quickly stop the spread of the disinformation through injunctive relief, and to seek damages as appropriate. There is no criminal liability.
AB 2655
The Limited Nature of AB 2655
AB 2655, effective Jan 1, 2025, seeks to stop, for a strictly limited time before and after elections, the online spread of the worst deepfakes and disinformation maliciously intended to prevent voters from voting or getting them to vote erroneously based on fraudulent content.
It does so by requiring large online platforms to remove such content upon notification. Content that may be banned is limited to deepfakes of candidates, elections officials, and elected officials intended to defraud voters. The bill also requires that other fake online content related to elections must be labeled as fake, again just for a limited time.
- It applies only to “materially deceptive” deepfake content, meaning it only applies to content that would deceive a reasonable person. Things that are obviously fake, jokey, cartoonish, or satirical are not touched by the bill. (§ 20513(a))
- It requires the content to be distributed with malice. (§§ 20513(a)(2)(C), 20514(a)(3))
- It only applies within 120 days before an election and, for those deepfakes regarding election officials, within 60 days after the election. It applies only when the risk of election disinformation is highest, not at all times. (§ 20513(e))
- Parody or satire are specifically exempted from the legislation. (§ 20519(c))
What Would AB 2655 Apply to, In Real Terms?
An example of a political deepfake the bill would apply to is a Tik Tok post that includes a deepfake of a local mayor “caught on tape” taking a bribe, right before Election Day, or an X post that includes a deepfake of a county elections official saying that their voting machines were all hacked, right after Election Day.
What are the Remedies in the Law?
AB 2655 is enforceable through an expedited process to quickly stop the spread of the disinformation through injunctive relief. There is no criminal or civil liability.