California Court Decision Misses the Mark on the Threat of Political Deepfakes

A federal district court in California has issued a preliminary injunction against a California state law, supported by Public Citizen, that aimed to curb deceptive AI-generated deepfakes that could influence the outcome of elections.

The court said that the law doesn’t take an adequately narrow approach to restricting such content, thereby infringing on First Amendment-protected speech or compelling unduly burdensome speech to avoid liability.

Robert Weissman, co-president of Public Citizen, issued the following statement:

“The court’s decision misses the fundamental problem with deepfakes, which is not simply that they make false claims but that they show candidates saying or doing things that the candidates did not say or do. The court suggests that a targeted candidate can just respond with counter speech – but that is not true, where the candidate has to ask the public not believe their eyes and ears. For this unprecedented kind of fraudulent deception, disclosure *is* the least speech-restrictive solution that can advance the government’s compelling interest in preserving election integrity.

“There are particular features of the California law – such as the requirement to label satire and font size required for disclosures – that appeared to color the court’s decision. And the court recognized that labeling requirements, if narrowly tailored, could pass constitutional muster. This decision therefore should not become an excuse for inaction against the threat of deepfakes.

“There’s nothing about the First Amendment that ties our hands in addressing fraud and a here-and-now threat to democracy.”