LOS ANGELES — Reality TV personality Spencer Pratt, a candidate for mayor of Los Angeles, has shared viral AI-generated campaign videos that cast him as Batman and Luke Skywalker in fan-made productions promoting his bid. The ads, created by AI filmmaker Charlie Curran, founder of Los Angeles-based Menace Studio, have circulated widely online and have received both praise and mockery.

The videos use generative AI tools to clone human voices and images, depicting Los Angeles under Democratic leadership as a city in distress and portraying Mayor Karen Bass and Governor Gavin Newsom as conspiring against residents. Pratt and representatives of his campaign could not immediately be reached for comment.

Bass responded to Pratt's online output. "His social media is now taking on a violent turn," she said.

Matt Klink, a Republican political consultant, defended the approach. "Spencer Pratt is using AI the way it should be used, which is to sharpen reality," Klink said. "His whole shtick is that Los Angeles is broken, the insiders have failed, and the political class wants to explain away what voters are seeing with their own eyes," he said.

The Pratt videos have emerged in a regulatory landscape that researchers describe as fragmented. No comprehensive federal rules govern the use of AI content in political ads. According to the National Conference of State Legislatures, 29 states have passed laws restricting deepfakes in political campaigns. Texas and Minnesota prohibit their use within a set window before an election, while the other 27 states require a disclosure when content contains a deepfake.

In September 2024, Newsom signed a package of bills clamping down on AI in politics. A federal judge later blocked one of the California laws regulating materially deceptive election-related content, ruling it likely violated the First Amendment.

Mark Jablonowski, chief executive of DSPolitical, said the spread of synthetic content places a burden on the public. "When you're creating content that is not based in reality, and then platforms are amplifying it in order to attract more eyeballs, you are putting a burden on the public for figuring out what is real and what is factual, and what is fake and misleading," he said.

In January 2024, Republicans placed robocalls using an AI deepfake voice mimicking President Biden, urging New Hampshire residents not to vote in the Democratic presidential primary. State authorities said the message violated voter suppression laws, and the company that sent the calls agreed to pay a $1 million fine. The following month, the Federal Communications Commission outlawed robocalls using AI-generated voices.