Three California AI bills are worth supporting as they face key tests this month

A handful of AI-related bills are making progress in the California legislature. If approved, they could establish a template for AI guardrails around the nation. Photo: Victoria Bernal.

In the emerging environment of AI regulatory policy, the Brookings Institute recently noted that “California is uniquely positioned to have a crucial impact on AI governance.”

That’s because 32 of the top 50 global AI firms—including OpenAI, Google, Anthropic, and Meta—are based in the Golden State. Congress has yet to pass a significant federal AI bill. So California’s state legislators currently have the chance to set the ground rules for the entire nation.

As of mid-June, a handful of significant bills addressing training data and AI transparency have been approved by their house of origin, effectively surviving the Sacramento session’s early culling.

TCAI’s high priority california bills

Three bills are showing particularly strong promise in terms of advancing the cause of transparency in the development and deployment of artificial intelligence systems. The Transparency Coalition’s Jai Jaisimha and Rob Eleveld will be in Sacramento in the coming weeks to offer expert support and help move these bills forward.

Key dates: These bills will be scrutinized over the next two weeks at two important hearings. On June 18, the Assembly Privacy and Consumer Protection Committee will consider SB 942 and SB 1047. On June 25, the Senate Judiciary Committee will look at AB 2013, AB 886, and AB 2877.

Here’s a brief overview of the top bills.

AB 2013: DAYLIGHTing THE TRAINING DATA

Assemblymember Irwin’s proposed Training Data Transparency Act represents one of the first serious efforts to enact safeguards around the data used to train AI models like OpenAI’s ChatGPT, Google’s Gemini chatbot, and Microsoft’s Copilot. Currently the use of that source material is unregulated, which means an AI model could learn equally from accurate information and harmful falsehoods. That can lead to the generation of biased or untrue outputs, hallucinations, and the perpetuation of damaging fabrications.

The bill aligns with our mission here at the Transparency Coalition: Move training data into daylight.

AB 2013 requires an AI model developer to post a high-level summary of the datasets used in the development of the AI system. That would include the sources or owners of the datasets, the number of data points in the datasets, whether the datasets were purchased or licensed by the developer, and whether the datasets include personal information.

AB 2013 passed the Assembly last month and goes before the Senate Judiciary Committee on June 25. TCAI will be there to support and cover the bill’s progress. A full analysis is available here.

AB 886: Licensing journalism used as training data

This is a promising proposal to protect the work and intellectual property rights of content creators. Assemblymember Buffy Wicks’ California Journalism Preservation Act would establish a licensing fee system paid by major tech platforms for the use of original journalism, including its use as training data for AI models.

Under AB 886, large online platforms would be required to enter into binding arbitration with digital journalism providers to establish an appropriate percentage of advertising revenue to be paid to those journalism providers for accessing their output. The bill defines “accessing content” to include acquiring, crawling, and indexing content. To insure that the increased revenue is reinvested in newsrooms, the bill requires that 70% of the funds derived from the bill must be spent on news journalists employed by the digital journalism provider.

AB 886 passed the Assembly on June 1. It next goes before the Senate Judiciary Committee on June 25. The Assembly analysis is available here.

SB 942: Required transparency: ‘this content is ai generated’

Senator Josh Becker’s California AI Transparency Act would require California’s largest generative AI developers to create an AI detection tool by which a person can query the covered provider as to the extent to which text, image, video, audio, or multimedia content was created, in whole or in part, by a generative AI system.

SB 942 would also mandate that covered developers affix a digital watermark to any AI-generated image, text, video, or multimedia content created by a generative AI system.

The Act also protects consumers by prohibiting AI developers or deployers from collecting, retaining, or revealing personal information from consumers who use the AI system to create AI-generated content submitted to the detection tool.

Limitations of the bill: SB 942 is aimed at the largest tech companies, and applies only to businesses that provide a generative AI system that has, on average over the preceding 12 months, over 1,000,000 monthly visitors or users and is publicly accessible within California’s geographic boundaries.

Approved by the Senate on May 21, the bill next goes before the Assembly’s Privacy and Consumer Protection Committee (P&CP) on June 18. The Senate analysis of SB 942 is available here.

two more worth watching

Two more bills are on our radar this month in Sacramento:

·      AB 2877: CCPA revision. Assemblymember Rebecca Bauer-Kahan’s bill would amend the California Consumer Protection Act to prohibit developers from collecting and using the personal information of consumers under 16 years of age to train an AI system. Passed in the Assembly on May 20, it will be heard by the Senate Judiciary Committee on June 25.

·      SB 1047: Safe and Secure Innovation for Frontier Artificial Intelligence Models Act. Sen. Scott Weiner’s ambitious AI regulatory package passed the Senate on May 21. It will be heard by the Assembly’s P&CP Committee on June 18.

Keep up with these and other AI transparency developments by signing up for our email connect below.