San Francisco Chronicle

Bay Area lawmakers push AI regulation

The California Capitol will likely see a flurry of attempts to regulate artificial intelligen­ce next year, with two Bay Area lawmakers emerging as leaders on the issue.

Assembly Member Rebecca Bauer-Kahan, D-Orinda, will play a key role in shaping any major AI legislatio­n in her new role as chair of the Assembly Privacy and Consumer Protection Committee. She says she’s focused on crafting bills tailored to address specific “known harms” from the technology, such as discrimina­tion and bias in algorithms and maintainin­g the privacy of medical records as doctors embrace AI in their practices.

She said that she expects all major AI legislatio­n to come before her committee, and that successful bills will need to show what specific problems they intend to address. She said legislatio­n needs to be tailored so that it doesn’t stifle innovation in the field or push smaller companies out of the industry.

“For any bills that come before the committee … we will be asking that basic fundamenta­l question, ‘What are the harms you’re trying to solve? ’ ” she said. “We’re going to want to see concrete harms, and we’re going to want to see feasible solutions.”

In that spirit, she said she plans to reintroduc­e her bill next year to ban companies from using AI-powered algorithms that discrimina­te against people. The bill would have required companies that develop those algorithms to assess them and document their intended uses, limitation­s and potential discrimina­tory risks. It was shelved in the Assembly Appropriat­ions Committee in May, but Bauer-Kahan says the new version will be “more streamline­d and allow for better compliance.”

Sen. Scott Wiener, D-San Francisco, has introduced legislatio­n that takes a broader approach. His bill, SB294, is now light on details but wide in scope, and intends to set industrywi­de safety and transparen­cy standards.

Wiener has not yet finalized the text of his bill, but announced it in September as an effort to “enact sweeping safety rules governing AI developmen­t” and characteri­zed it as “among the first attempts at broad regulation of AI.”

The legislatio­n is focused on a number of potential risks posed by AI, including its potential to be used to create biological weapons, target critical infrastruc­ture and spread disinforma­tion.

“To be clear, I’m not an AI skeptic,” Wiener told the Chronicle on Wednesday. “AI has huge potential benefits for humanity. And we need to foster those benefits while we assess and then mitigate risks.”

Despite his bill’s scope, he said that it doesn’t address all potential AI regulation, and that he expects to see a significan­t number of other AI bills introduced.

Chandler Morse, vice president of corporate affairs for the finance and HR software company Workday, said he’s never seen such a dramatic increase in an area of policy as he has in AI in recent months. Workday is among the companies working to shape AI regulation at the state, national and internatio­nal levels. Morse pointed to recent efforts in Washington, Connecticu­t, Maryland and New York to enact AI regulation, but also said he thinks California is the furthest along because of Bauer-Kahan’s plan to reintroduc­e her algorithm bias bill.

“If we can successful­ly land a workable bill in California, I think it will help as other states start to think about how to do this,” he said.

Workday and other companies seeking to shape the regulation­s that will govern the AI industry have also pushed for the federal government to enact a uniform policy for the country.

President Joe Biden signed an executive order in October that requires companies building advanced AI models to test their safety and directs federal agencies to monitor their use of AI for risks. But gridlocked Congress has not passed major regulation of the technology.

Bauer-Kahan said she would also like to see Congress act, but she has low expectatio­ns.

“A consistent regulatory framework across the country is the best thing for consumers and for business, but I think we’re all very aware that’s not going to happen,” she said. “They seem to not be able to pass a budget, so I don’t believe they’ll pass AI legislatio­n.”

Instead, she said, she hopes California leads on legislatio­n that other states will replicate. That’s already happening with her AI discrimina­tion bill, she said.

In addition to Wiener and Bauer-Kahan’s bills, Assembly Member Ash Kalra, D-San Jose, has introduced a measure to limit the ability of movie studios and other entertainm­ent companies to use AI-generated versions of human performers in production­s. Jonathan Mehta Stein, who leads the voting rights advocacy organizati­on California Common Cause, said the group plans to propose several bills focused on regulating the use of AI to influence elections early next year.

Bauer-Kahan said she sees part of her role as assessing existing laws — such as those banning child pornograph­y, protecting intellectu­al property and prohibitin­g discrimina­tion — and ensuring that they will apply to artificial intelligen­ce.

“What we also need to do is not necessaril­y come up with new laws, but make sure that the laws we have on the books are protective of the situation where your actor has changed from human actors to an artificial intelligen­ce tool,” she said. “How do we make sure that our laws continue to protect the people we’ve been protecting?”