California lawmakers discuss protecting actors, creators from AI exploitation
Artists, lawmakers and artificial intelligence collided this week during a hearing of two legislative committees that drew a standing-room-only crowd.
At issue was copyright law and how it intersects with artificial intelligence, an issue that’s been gaining prominence since Scarlett Johansen called out OpenAI in May 2024 for giving ChatGPT a voice that sounded remarkably like hers.
The Monday hearing at Stanford University brought together the Senate Judiciary Committee and Assembly Privacy and Consumer Protection Committee. According to privacy committee chair Assemblymember Rebecca Bauer-Kahan, D-Orinda, the idea for the hearing stemmed from a bill that was introduced during the last legislative session.
That bill, Assembly Bill 412, would force artificial intelligence companies to be more transparent about what copyrighted materials were used to train the system or model. The bill was authored by Bauer-Kahan and stalled in the State Senate, but is expected to be taken up again during the next legislative session.
Particularly instructive on the copyright issue, Bauer-Kahan said, was the testimony from Julian Brabon, the CEO and co-founder of Stability, a company that allows artists to check to see if their work has been used to train AI models. During the hearing, Brabon did a live demonstration, showing that various artworks and even photos of Umberg and Bauer-Kahan had been used to train AI models.
“Often people want to say that this is a black box and nobody can know,” Bauer-Kahan said in an interview Tuesday. “He really proved that wasn’t the case.”
Lawmakers also heard from a representative from Open AI, which the Orinda assemblywoman said was the only Large Language Model company to send a representative.
Mark Gray, who serves as a copyright policy attorney for the company, stressed that the company already has guardrails to prevent replication of someone’s likeness, and wondered if copyright was the right “lever” to pull to preserve the creative economy. He said the company was proud of its existing partnerships with news organizations and some creative sectors.
“At the end of the day the question really is just whether the training level data is the right layer to do it,” he said.
The legislators present urged the hearing’s attendees to present them with creative solutions for protecting the state’s creative economy. Bauer-Kahan said in an interview after the fact that she hopes more of California’s more than 30 AI companies will come to the table to discuss the future of copyright law.
“There’s no question that we can’t just allow our creative economy and our artists...to not be protected in this moment of transition,” she said. “Ideally we’d like to do that together.”
