Key California Lawmaker Unveils AI Anti-Discrimination Measure

A key California lawmaker filed highly anticipated legislation late Thursday that would prohibit discrimination from AI software, as businesses increasingly turn to the emerging technology and automation to make hiring and other important decisions.

Assemblymember Rebecca Bauer-Kahan (D) revamped her bill after her previous legislation died last year due to state budget concerns and opposition from tech groups. Her new measure (A.B. 2930) though could still encounter those same problems in a bid toward passage this year.

Bauer-Kahan’s initial bill last year was the only comprehensive state-level AI legislation then, and it inspired similar measures this year in other states such as Washington and Connecticut. Analysts are watching to see if the new bill may have even bigger policy influence, especially after Bauer-Kahan scrapped language that would have allowed state residents to sue over violations.

“We’re kind of looking for California to be a leader in this space and a model that can be replicated in other states,” said Chandler Morse, vice president of public policy for Workday, the California-based company that provides human resource software to businesses.

New Changes

Bauer-Kahan negotiated with various companies in the tech space to refine the measure that would curb algorithmic discrimination. The new bill arguably has more support than last year from such companies like Workday, which provided input to the lawmaker.

“Our AI policy has been around supporting regulatory approaches that can build trust and support innovation,” said Morse. “There were a couple pieces of it, particularly around the enforcement issues that we were kind of hoping to see changed.”

The lobbying has resulted in changes that may appease some concerns from businesses. Most notably, she dropped language for a private right of action that would have let Californians sue businesses for discrimination.

Only the state attorney general and public attorneys would be able to file suit under the new bill. The state Civil Rights Department would also have investigation powers.

Many of the features of last year’s bill have remained, such as the focus on AI tools making “consequential” decisions in areas like housing or employment. Such tools would have to go through impact assessments, where companies evaluate the pros and cons of using such technology. The measure, like last year’s, makes a distinction between businesses that develop versus those that deploy automated systems and gives them different compliance responsibilities.

Bauer-Kahan’s new leadership position will enhance the measure’s chances. She now heads the California Assembly’s privacy committee, which many AI-related pieces of legislation will have to pass through.

Same Hurdles

Not everybody in the tech sector will be satisfied, a fact that Bauer-Kahan has acknowledged.

“The technology companies, obviously, I think there’s some pushback. Not all want to be regulated,” she said in an interview last month. “Part of what we’ve been trying to do is really bring those conversations to the table to try to figure out how we meet in the middle. That’s really hard.”

Tech groups are likely to argue again that the bill is too broad or places too many restrictions. Adam Kovacevich, CEO of tech trade group Chamber of Progress, said in an interview before the bill introduction that the key thing he’s looking for is whether the measure will include a safe-harbor provision for unintentional discrimination caused by algorithms.

“No AI innovator wants their service to be used to perpetuate discrimination,” he said. His group needs to review the bill first before taking a formal position.

California also is still facing a state budget deficit of $37 billion, a reason last year’s measure failed. The legislation calls for more resources for enforcement. However, the ever-increasing urgency to address AI’s negative impacts—as evidenced by the numerous AI bills filed this year compared to last year—may help spur passage.

“Algorithmic discrimination is going to continue to be a priority for me,” said Bauer-Kahan. “I think it’s one of the top things when we talk about harms from AI.”