San Mateo County lawmakers address AI concerns

Assembly Bill 1831, authored by Berman, would outlaw the creation, distribution and possession of artificial intelligence-generated child sexual abuse material. Berman began working on the issue after being approached by law enforcement who informed him that current law prohibiting the creation and possession of child pornography only applies to content of actual children, not fake images crafted using images scrubbed from the internet. 

Citing a report by the Internet Watch Foundation, Berman said thousands of highly realistic explicit images of children are being produced using AI and published on one dark web forum in a one-month period. 

“This is a really disturbingly large problem. … It’s really important this loophole is closed,” Berman said. “When I first heard about it, I couldn’t believe it was a thing. Once you know, you’ve got to do something about it and that’s what we’re going to do.” 

Becker’s bill, dubbed the California Artificial Intelligence Transparency Act, would require large AI companies to watermark images, videos and audio created through their services and to create a platform where consumers can ask a company if they’re behind a piece of content. 

The idea for the bill was sparked after Becker learned students were being accused of using AI to write essays and were unable to prove otherwise. Before introducing the bill, Becker said he was also familiar with scammers using snips of peoples’ voices to dupe others into sharing personal information. 

Both Becker and Berman said they intend to use the legislative season to improve their respective bills. Becker acknowledged that his bill only applies to larger AI companies while Berman said his team is looking into whether AI companies can be held responsible for their platforms being used to create child sexual abuse material. 

“We’ve got to start somewhere and that’s really the goal of this bill,” Becker said. “The technology is moving so fast and the danger is we do something irrelevant because the technology moved past it or we do something misguided because we don’t understand the technology. Those are the risks and I’m conscious of that.” 

Becker said he’s been in conversations with the Stanford University’s Institute for Human-Centered AI, AI companies and advocates on his bill and Berman shared confidence that AI providers would be interested in doing their part to protect the public. 

The legislators also said they hope their legislation encourages people to be more cautious about the type of content they share and consume online and in their day-to-day lives. Both have also been behind other recently enacted laws related to media literacy and privacy laws including Becker’s Delete Act, which allows consumers to opt out of having their data collected and sold and Berman’s AB 873, which requires media literacy courses to be taught to students in kindergarten to 12th grade. 

“We all have a responsibility, a very, very serious responsibility, to anticipate how artificial intelligence technology might be abused and to mitigate that before it happens,” Berman said. “I hope it leads to a greater awareness of the risks of putting any images online … and how those images can be stolen and manipulated and abused for terrible purposes.”