Last summer, Paul Lehmann and Linnea Sage talked about the rise of artificial intelligence and the threat it poses to the lives of writers, actors, and other entertainment professionals as they drove to a hospital near their Manhattan home. I was listening to the podcast.
This topic was especially important for young couples. They made a living as voice actors, and AI technology was beginning to generate lifelike voices.
But the podcast took an unexpected turn. To highlight the threat of AI, the hosts conducted a lengthy interview with a conversational chatbot named Poe. It was like Lehman.
“He was interviewing me for my voice about the dangers of AI and the harm it could do to the entertainment industry,” Lehmann said. “We pulled over and sat there in complete disbelief, trying to understand what had happened and what we should do.”
Lehmann and Sage are currently suing the company that created the bot's audio. They allege that Lovo, a startup in Berkeley, Calif., illegally used the audio recordings to develop technology that could compete with the audio works. After hearing Lehmann's voice cloned on a podcast, the couple realized that Lobo had also cloned Sage's voice.
The couple joins a growing number of artists, publishers, computer programmers, and other creators suing developers of AI technology, and that these companies will ultimately take their chances as they create tools to replace AI in the job market. They allege that the work was used without permission and are joining in. (The New York Times sued two companies in December, accusing OpenAI and its partner Microsoft of using the company's copyrighted news articles to build online chatbots.)
In a lawsuit filed Thursday in federal court in Manhattan, the couple alleges that an anonymous Lovo employee paid for several audio clips in 2019 and 2020 without disclosing how the clips were used. said.
They allege that Lovo, which was founded in 2019, violates federal trademark law and several state privacy laws by promoting voice clones. The lawsuit is seeking class action status, and Lehmann and Sage are inviting other voice actors to join.
“We don't know how many other people have been affected,” their attorney Steve Cohen said.
Lobo denies the allegations in the lawsuit, said David Case, an attorney representing the company. He added that “there would be no problem” as long as all individuals providing voice recordings to Robo consented.
The company's CEO, Tom Lee, said on a podcast episode last year that Lovo currently offers revenue that allows voice actors to help create voice clones for the company and receive a portion of the revenue from those clones. He said he is offering a distribution program.
Jeffrey Bennett, general counsel for SAG-AFTRA, the union representing 160,000 media workers around the world, said the lawsuit appears to be the first of its kind.
“This case will show people, especially technology companies, that your voice has a right and that there are people out there who make a living using their voice,” he said. Ta.
In 2019, Lehman and Sage advertised themselves as voice actors on Fiverr, a website that allows freelance professionals to promote their work. Through this online marketplace, she was often asked to provide voice work for commercials, radio ads, online videos, video games, and other media.
That same year, Sage was contacted by an anonymous person who paid him $400 to record several radio scripts, with the explanation that the recordings would not be used for public purposes, according to correspondence cited in the complaint. It is described in.
“These are test scripts for radio ads,” the unnamed person said, according to the complaint. “They are not exposed externally and are only used internally, so they do not require rights of any kind.”
Seven months later, another unidentified person contacted Lehman about a similar job. Rehman, who also works as a TV and film actor, asked how the clips would be used. According to communications cited in the complaint, the person stated multiple times that it would be used only for research and academic purposes. Mr. Lehmann was paid $1,200. (He provided a longer recording than Mr. Sage.)
In April 2022, Lehmann discovered a YouTube video about the Ukraine war narrated by a voice similar to his own.
“This is my voice talking about weapons in the Ukraine-Russia conflict,” he said. “I went blank and had goosebumps on my arms. I knew I had never said those words in that order.”
For months, he and Sage struggled to understand what had happened. They hired a lawyer to track down who made the YouTube video and how Lehmann's voice was reproduced. However, the owner of the YouTube channel appears to be based in Indonesia, and there was no way to find him.
Then they listened to a podcast on the way to the doctor's office. Through the podcast Deadline Strike Talk, they were able to determine the source of Lehmann's voice clone. A professor at the Massachusetts Institute of Technology built a chatbot using Lovo's speech synthesis technology.
Sage also found an online video of the company pitching its voice technology to investors during an event in Berkeley in early 2020. In the video, Lovo executives showed a synthesized version of Sage's voice and compared it to the audio. She recorded her real voice. Both performed alongside photos of women who were not hers.
“I appeared in their pitch video to raise money,” Sage said. The company has since raised more than $7 million and claims to have more than 2 million customers worldwide.
Ms. Lehman and Ms. Sage also discovered that Lovo was advertising voice clones of her and Mr. Lehman on her website. After they sent the company a cease-and-desist letter, the company announced it had removed the voice clones from its site. But Lehmann and Sage argued that the software powering these voice clones has already been downloaded by countless of their customers and can still be used.
Lehmann also questioned whether the company used the couple's voices, along with many others, to build the core technology that powers its voice cloning system. Just as OpenAI's ChatGPT and other chatbots learn skills by analyzing vast amounts of text culled from the internet, speech synthesizers learn skills by analyzing thousands of hours of spoken language. Often learn.
According to court documents, Lovo admitted to training its technology by recording thousands of hours of audio.
Case, the attorney representing Lovo, said the company trained its AI system using audio from a freely available database of English recordings called Openslr.org. He did not respond to questions about whether audio recordings of Mr. Lehmann and Mr. Sage were used for technical training.
“We want to take back control over our voices, ourselves and our careers,” Lehman said. “We want to represent the people who have happened and what will happen if nothing changes.”