According to Friday complaints, David Millette said Google's Gemini and OpenAI's ChatGPT large language model programs are being unlawfully "trained" with unauthorized transcriptions of people's YouTube videos that are allegedly being fed into them by the companies in order to make those programs more capable of imitating human speech patterns in its replies to user queries. Millette said the companies have unfairly enriched themselves with their alleged "surreptitious, non-consensual transcription and use of millions of YouTube users' videos."
"For years, YouTube has been a popular video sharing platform that allows content creators and users to upload and share videos with audiences worldwide," the largely identical complaints state. "However, unbeknownst to those who upload videos to YouTube, defendants have been covertly transcribing YouTube videos to create training datasets that defendants then use to train their AI products."
Millette seeks to represent two nationwide classes, one for the case against Google LLC, Alphabet Inc. and YouTube Inc. and another for the suit against OpenAI Inc. and several of its affiliated entities. The proposed nationwide classes would include everyone living in the U.S. whose YouTube videos were transcribed and then used as "training data" for the companies' AI products without their consent.
In addition to the nationwide classes, Millette also seeks to establish a subclass in each state for California residents like himself. The companies as well as their affiliates, workers and any parents or children of those workers are excluded from the proposed classes, the suits say.
Google and OpenAI both face claims for unjust enrichment and unfair competition, with the former claim being brought under federal law and the latter coming under the California Business and Professions Code.
According to the complaints, Google and OpenAI have profited from training their AI products using the proposed class members' words and ideas without permission. Those AI products, which function by generating imitations of human speech and artistic works based on the data fed into them, are currently ripping off and competing with the video creators they are taking those words and ideas from, the suits say.
Millette's suits seek injunctive relief blocking the companies from transcribing the proposed class members' videos to train their AI products without consent as well as compensatory, statutory and punitive damages with interest along with attorney fees and costs.
Counsel for Millette and representatives for the companies did not immediately respond to requests for comment Monday.
Millette is represented in both cases by L. Timothy Fisher, Joshua B. Glatt, Joseph I. Marchese and Julian C. Diamond of Bursor & Fisher PA.
Counsel information for Google and OpenAI was not immediately available.
The cases are Millette v. OpenAI, Inc. et al. and Millette v. Google LLC et al., case numbers 3:24-cv-04710 and 5:24-cv-04708, both in the U.S. District Court for the Northern District of California.
--Editing by Emily Kokoll.
For a reprint of this article, please contact reprints@law360.com.