Mark Zuckerberg doesn’t really want to answer tough questions about Meta’s artificial intelligence initiatives and goals, but a federal judge told the Facebook founder this week that’s exactly what he has to do.
“Plaintiffs have presented evidence showing that Zuckerberg was a key decision maker and policy maker in the development of Meta’s generative AI division and the large-scale language models at issue in this litigation,” U.S. District Judge Thomas Hixson said on September 24 about the potential class action lawsuit first filed last year by Sarah Silverman, Richard Kadry, and Christopher Goldenam, and now joined by Ta-Nehisi Coates and others.
In addition to the more critical lawsuit against OpenAI, authors sued Meta in mid-2023 over copyright infringement concerns that their works and books were illegally downloaded and used to train the company’s large-scale language model AI software.
Bedwetter authors Silverman and National Book Award winner Coates, along with the other plaintiffs, allege that “much of the material contained in Meta’s training datasets comes from copyrighted works, including books written by Plaintiffs, and that Meta has used, without consent,
No credit, no compensation.”
Despite the legal leeway, Meta denies that it accessed authors’ copyrighted work because of its LLaMA system. Meta’s lawyers are also trying to push the argument that the tech giant has many people better suited than Zuckerberg to be cross-examined by David Boies and other plaintiffs’ lawyers.
It didn’t fly.
“Plaintiffs do not generally allege, as Meta suggests, that Mr. Zuckerberg is responsible for everything because he is the CEO of the company,” the judge wrote in his order denying Meta’s motion to block the CEO from facing cross-examination by Silverman and other lawyers. “Rather, plaintiffs have presented evidence of his specific involvement in the company’s AI initiatives. Plaintiffs have presented evidence showing that Mr. Zuckerberg was the primary decision maker regarding Meta’s decision to open-source its language model. Plaintiffs have also presented evidence that Mr. Zuckerberg had direct oversight of Meta’s AI products.”
Judge Hixson also stated, “Given this factual showing, the Court does not intend to require Plaintiffs to exhaust other forms of discovery before calling Mr. Zuckerberg to testify. Plaintiffs have made a strong case that this deposition is meritorious.”
No date has yet been set for a deposition for Zuckerberg, who doesn’t particularly like being in front of a microphone. A discovery hearing on the matter just wrapped up this afternoon in San Francisco, so the deposition could happen sooner rather than later.
By then, everything about AI may have changed again.
It’s been almost two years since ChatGPT brought AI to the masses, and the technology is quickly coming to the forefront in almost every aspect of society and industry.
Results vary depending on the perspective.
For example, California Governor Gavin Newsom signed a bill earlier this month that provides some protection for the public image rights of actors, performers, living people and business associates. At roughly the same time, Lionsgate and applied AI research company Runway announced a partnership on September 18 to develop AI customized for the studio’s unique portfolio of film and TV content, including John Wick.
Twerking at the courts and pushing for the inevitable future, Zuckerberg took to the stage at the company’s Meta Connects conference in Menlo Park, California to talk all things AI, including the news that Meta’s AI chatbots will soon be able to communicate in the voices of Awkwafina, Judi Dench, Kristen Bell, John Cena and Keegan-Michael Key.
Unfortunately, Zuckerberg will have to testify in his own voice.
