California Bills Would Impact Artificial Intelligence, Shape Security

Movement on AI, Other Measures in California’s 2024 Legislative Session

As California’s legislature enters the last days of its legislative session that ends Aug. 31, 2024, it’s do-or-die time for key legislation on artificial intelligence (AI) that has been discussed and debated over for months.

The Senate and Assembly Appropriations Committees faced an Aug. 15 deadline to either hold or clear bills for floor consideration that are determined to have a significant fiscal impact on the state. Bills held at this point will not see further action this session.

Several key AI bills that the Security Industry Association (SIA) and other business and industry organizations are tracking closely were cleared for floor consideration by the Appropriation Committees – some with very significant amendments as a condition for moving the measures to the next stage:

AB-2930: Automated Decision Tools

This bill, introduced by California Assemblymember Rebecca Bauer-Kahan (D-Orinda), would impose requirements on developers and deployers of “automated decision tools” including restrictions on “algorithmic discrimination.” The latest committee amendments dramatically scale back the applicability of the measure by limiting it to employment related decisions, exempting state and local agencies and limiting enforcement authority to the California Civil Rights Department. Business community concerns remain however regarding an overall framework of related definitions and regulatory mechanisms that are not scaled based on risk, the scope of which could be dramatically expanded in the future to apply to more technologies.

SB-1047: Safe and Secure Innovation for Frontier Artificial Intelligence Models Act

This measure, introduced by California State Sen. Scott Wiener (D-San Francisco), would establish state standards and regulatory requirements applicable to developers and users of generative AI technology, and has been the focus of tech industry concerns as well as criticism from former U.S. House speaker Rep. Nancy Pelosi (D-Calif.) and others.

SB 892: Public Contracts, Automated Decision Systems, Procurement Standards

This measure, introduced by California State Sen. Steve Padilla (D-San Diego), establishes stringent new assessments and controls for state agencies using “automated decision systems,” as it very broadly defines as any “computational process” including AI that “is used to assist or replace human discretionary decision making and materially impacts natural persons.”

SB-896: Generative Artificial Intelligence Accountability Act (Passed House Appropriations)

SB-896, introduced by California State Sen. Bill Dodd (D-Napa), directs the California Department of Technology and other agencies to investigate and and update guidance to agencies on guidance for using generative AI. 

Another AI measure of interest was not subject to Appropriations Committees and is receiving floor consideration with amendments in the coming two weeks:

AB-2013: Transparency on AI Training Data

This measure, introduced by California Assemblymember Jacqui Irwin (D-Thousand Oaks), requires AI developers to publicly disclose certain information about training data used. It was recently amended to apply only to generative AI systems. Earlier in the process, a clarification was added that excludes AI used for cyber and physical security purposes from these regiments to address the potential risk of compromise to such systems through disclosure, a key outcome actively encouraged by SIA.

Measure Held in Appropriations:

AB-814: Law Enforcement Use of Facial Recognition

Endorsed and actively supported by SIA, this measure, introduced by California Assemblymember Phil Ting (D-San Francisco), would establish foundational safeguards for use of facial recognition technology by law enforcement agencies by ensuring potential match results are never used as the sole basis for probable cause for an arrest or search. The measure was held despite unanimous passage by four different House and Senate committees and a unanimous bipartisan vote in the full assembly in May. Activist groups advocating a ban on the technology had earlier mounted an organized campaign of opposition to the measure.

California saw a large number of AI-related bills introduced for consideration this year compared to other states, raising concerns about the risk of haphazard drafting and misalignment among them that could result in unintended consequences and future compliance difficulties. It is likely that California Gov. Gavin Newsom, whose administration has already weighed in on some of these measures, might have concerns similar to those expressed by governors in other states regarding high stakes involved and potential for overreach, considering the impact this could have on local AI developers and industry sectors that have been a critical source of economic growth.

SIA will provide an update on the final outcome on these measures as the the 2024 session comes to a close and Gov. Newsom considers legislation approved by the legislature.