California 2025-2026 Regular Session

California Senate Bill SB53

Introduced
1/7/25  
Refer
1/29/25  
Refer
2/27/25  
Refer
3/12/25  
Report Pass
3/26/25  
Refer
3/27/25  
Report Pass
3/26/25  
Report Pass
4/9/25  
Refer
3/27/25  
Refer
4/9/25  
Report Pass
5/23/25  
Report Pass
4/9/25  
Refer
4/9/25  
Engrossed
5/28/25  
Report Pass
5/23/25  
Report Pass
5/23/25  
Refer
6/5/25  
Engrossed
5/28/25  
Refer
6/12/25  
Refer
6/5/25  
Refer
6/12/25  
Report Pass
7/2/25  
Refer
7/2/25  
Refer
7/8/25  
Report Pass
7/16/25  

Caption

Artificial intelligence models: large developers.

Impact

The TFAIA introduces significant changes to existing laws surrounding AI and emphasizes the need for accountability among large AI developers. It requires annual audits by independent third parties, who will assess compliance with the specified safety guidelines. Failure to comply with the new regulations will result in civil penalties enforced by the Attorney General. Additionally, the act aims to create a consortium to develop a public cloud computing framework, known as CalCompute, which will facilitate ethical and equitable AI development. This initiative is expected to greatly influence the regulatory landscape concerning AI technologies in California.

Summary

Senate Bill 53, titled the Transparency in Frontier Artificial Intelligence Act (TFAIA), is designed to enhance the transparency and safety of artificial intelligence (AI) systems developed by large companies. The bill mandates that these developers create and publish detailed safety protocols that outline their procedures for assessing risks associated with their AI models. Specifically, it aims to establish better oversight and accountability in the rapidly evolving field of AI technology. The act stipulates that developers must inform the public and authorities about the data used to train their models and the testing methods utilized to mitigate potential catastrophic risks.

Sentiment

The sentiment surrounding SB 53 is generally positive among advocates of transparency and safety in AI development. Proponents argue that the bill is a vital step toward ensuring that AI technologies do not pose risks to the public. However, there are concerns, especially from some industry stakeholders, regarding the potential burdens placed on developers and the possible stifling of innovation. Critics have raised questions about the extent of regulation and whether it could inadvertently hinder smaller developers or create barriers to entry into the AI sector.

Contention

One notable point of contention relates to the bill's impact on the balance between innovation and regulation. While the act aims to enhance safety, some industry supporters worry that the comprehensive compliance measures may be excessive, particularly for smaller developers. There is also a discussion surrounding the need for effective whistleblower protections, ensuring that employees can report safety incidents without fear of repercussions. The act contains provisions to protect whistleblowers, making it an essential aspect of the proposed changes.

Companion Bills

No companion bills found.

Previously Filed As

CA SB1047

Safe and Secure Innovation for Frontier Artificial Intelligence Models Act.

CA AB2885

Artificial intelligence.

CA AB3050

Artificial intelligence.

CA SB896

Generative Artificial Intelligence Accountability Act.

CA SB1288

Public schools: artificial intelligence working group.

CA AJR6

Artificial intelligence.

CA AB2811

Attorneys: court filings: artificial intelligence.

CA AB2652

State Department of Education: artificial intelligence working group.

CA SB970

Artificial intelligence technology.

CA SB1235

Public postsecondary education: Artificial Intelligence and Deepfake Working Group.

Similar Bills

CA SB1047

Safe and Secure Innovation for Frontier Artificial Intelligence Models Act.

CA SB294

Health care coverage: independent medical review.

IL HB3506

ARTIFICIAL INTELLIGENCE SAFETY

CA AB594

Artificial intelligence.

NY S06953

Relates to the training and use of artificial intelligence frontier models; defines terms; establishes remedies for violations.

NY A06453

Relates to the training and use of artificial intelligence frontier models; defines terms; establishes remedies for violations.

CA SB833

Critical infrastructure: artificial intelligence systems: human oversight.