Berkeley Startup Accused of Stealing Actors’ Voices to Create AI Generator

Lovo, a Berkeley-based AI startup, was accused in a class action lawsuit on Thursday of stealing actors’ voices to create Genny, a publicly accessible AI generator.

Paul Skye Lehrman and Linnea Sage, a New York couple who work as voice actors, alleged that they were deceived into providing voice samples through a freelance website. They were paid $1200 and $400, respectively, for reading scripts, and were told the recordings would be used internally, or for academic purposes.

More from Variety

Years later, they discovered that their voices had been cloned and were being used publicly without their permission or additional payment, the suit alleges. They are seeking to represent every actor whose voice was taken, and to stop Lovo from using those voices without consent and payment.

“You gotta get back control of their voices and their professional destiny,” said Steve Cohen, a partner at Pollock Cohen LLP, who filed the complaint. “There’s gotta be compensation for every actor whose voice was used.”

Lovo charges a monthly subscription fee to use its AI voice generator, which can be used to make ads, corporate training videos, product demos, or for customer service applications.

The company also advertises that its tool can be used to mimic celebrity voices, though only “for personal entertainment purposes.” The site features demos meant to sound like Ariana Grande, Barack Obama, Conan O’Brien and others.

Lovo was founded in 2019 by two UC Berkeley grads and had raised $7 million in venture capital as of January 2023.

According to the lawsuit, Lehrman got an inquiry in May 2020 on Fiverr, a freelance platform for creative services. The requester — identified only as User25199087 — told him that the audio files would be used for “academic research purposes only.”

“We are researching speech synthesis with different accents and voices,” the user stated, according to the complaint. The person declined to provide further explanation, saying the research was “a confidential work in process sorry haha.”

Two years later, Lehrman found out that his voice had been used to narrate a YouTube video about Russian military equipment. In June 2023, Lehrman heard himself being interviewed as an AI-generated character on an episode of the Deadline “Strike Talk” podcast.

“Mr. Lehrman never gave permission to LOVO or to the Deadline Strike Talk podcast to use his voice, nor was he ever compensated for that unauthorized use,” the lawsuit states.

Sage, likewise, was hired off the Fiverr platform to read radio scripts in October 2019. The requester was identified as “tomlsg,” whom the plaintiffs believe is Lovo co-founder Tom Lee. She was told the clips would be “test scripts for radio ads” and would not be disclosed externally.

In 2023, according to the suit, she discovered that her voice had been used in Lovo’s video pitch for investors. The suit alleges that Lehrman and Sage’s voices were used to create AI clones, identified on Lovo’s site as “Kyle Snow” and “Sally Coleman,” which users could select for their projects.

Their voices were ultimately removed from the site last fall after their lawyer complained, but the suit alleges that those voices are still available to anyone who had previously downloaded them.

The suit accuses Lovo of engaging in fraud, and of violating a New York law that forbids the commercial use of a person’s voice without consent. The suit also accuses Lovo of false advertising, and of violating the federal Lanham Act.

The estate of George Carlin filed a similar suit earlier this year under the Copyright Act and California’s right of publicity law, after a podcast used an imitation of Carlin’s voice to create an AI standup set.

In that case, the podcast hosts agreed to delete the video and accepted an injunction barring them from uploading it again or otherwise using Carlin’s voice or likeness on any platform.

In a podcast interview last year, Lee said he was “democratizing access to voices” by opening up the field to more actors.

“We’re not just taking voices for free,” he said. “We’re giving them fair value. We’re also doing revenue share with certain voice actors.”

He was also asked about the ethical considerations surrounding synthetic voice technology.

“Tech can be good or bad,” he said. “It really depends on how the person who’s wielding it uses it.”

Best of Variety

Sign up for Variety’s Newsletter. For the latest news, follow us on Facebook, Twitter, and Instagram.