2026-10-20 SB-53 Senate Governmental Organization analysis

Markdown copy of 2026-10-20 SB-53 Senate Governmental Organization analysis.pdf, made 2026-10-20. The original is the authority. SENATE COMMITTEE ON GOVERNMENTAL ORGANIZATION Senator Steve Padilla Chair 2025 - 2026 Regular

Bill No: SB 53 Author: Version: Urgency: Consultant: Brian Duke

Wiener 2/27/2025 Amended No

Hearing Date: 3/25/2025

Fiscal:

Yes

SUBJECT: CalCompute: foundation models: whistleblowers

DIGEST: This bill establishes, within the Government Operations Agency (GovOps), a consortium to develop a framework for the creation of a public cloud computing cluster to be known as “CalCompute,” as specified. Additionally, this bill enhances whistleblower protections for employees involved with covered artificial intelligence (AI) models and covered model derivatives, as specified.

ANALYSIS:

Existing law:

1. Establishes the California Department of Technology (CDT), within GovOps, and requires the department to conduct a comprehensive inventory of all high- risk automated decision systems that have been proposed for use, development, or procurement by, or are being used, developed, or procured by, any state agency, as specified.

2. The Generative Artificial Intelligence (GenAI) Accountability Act requires

CDT, under the guidance of GovOps, the Office of Data and Innovation (ODI), and the Department of Human Resources (CalHR), to update, as needed, the State of California Benefits and Risks of GenAI Intelligence Report.

3. Defines “artificial intelligence” to mean an engineered or machine-based

system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments.

4. Prohibits employers and their agents from making, adopting, or enforcing a

rule, regulation, or policy preventing an employee from disclosing information to certain entities or from providing information to, or testifying before, any

SB 53 (Wiener)

Page 2 of 11

public body conducting an investigation, hearing, or inquiry if the employee has reasonable cause to believe that the information discloses a violation of law, as specified, and prohibits retaliation against an employee for, among other things, exercising these rights.

5. Provides that any report required or requested by law to be submitted by a state or local agency to a committee of the Legislature or the Members of either house of the Legislature generally, to instead be submitted as a printed copy to the Secretary of the Senate, as an electronic copy to the Chief Clerk of the Assembly, and as an electronic or printed copy to the Legislative Counsel, as specified.

This bill:

1. Establishes a consortium, within GovOps, and requires the consortium to

develop a framework for the creation of a public cloud computing cluster to be known as “CalCompute” that advances the development and deployment of AI that is safe, ethical, equitable, and sustainable by doing, at a minimum, both of the following:

a. Fostering research and innovation that benefits the public. b. Enabling equitable innovation by expanding access to computational

resources.

2. Requires the consortium to make reasonable efforts to ensure that CalCompute is established within the University of California (UC) to the extent possible.

3. Requires CalCompute to include, but not be limited to, all of the following:

a. A fully owned and hosted cloud platform. b. Necessary human expertise to operate and maintain the platform. c. Necessary human expertise to support, train, and facilitate the use of

CalCompute.

4. Requires the consortium to operate in accordance with all relevant labor and

workforce laws and standards.

5. Requires GovOps, on or before January 1, 2027, to submit a report from the consortium to the Legislature with the framework developed pursuant to the above provisions for the creation and operation of CalCompute.

SB 53 (Wiener)

Page 3 of 11

6. Requires the report to include all of the following elements:

a. A landscape analysis of California’s current public, private, and

nonprofit cloud computing platform infrastructure.

b. An analysis of the cost to the state to build and maintain CalCompute

and recommendations for potential funding sources.

c. Recommendations for the governance structure and ongoing operation

of CalCompute.

d. Recommendations for the parameters for use of CalCompute,

including, but not limited to, a process for determining which users and projects will be supported by CalCompute.

e. An analysis of the state’s technology workforce and recommendations for equitable pathways to strengthen the workforce, including the role of CalCompute.

f. A detailed description of any proposed partnerships, contracts, or

licensing agreements with nongovernmental entities, including, but not limited to, technology-based companies, that demonstrates compliance with the requirements above.

g. Recommendations regarding how the creation and ongoing

management of CalCompute can prioritize the use of the current public sector workforce.

7. Requires the consortium to, consistent with state constitutional law, consist of

14 members selected from among all of the following:

a. Representatives of the UC and other public and private academic

research institutions and national laboratories.

b. Representatives of impacted workforce labor organizations. c. Representatives of stakeholder groups with relevant expertise and experience, including, but not limited to, ethicists, consumer rights advocates, and other public interest advocates.

d. Experts in technology and AI to provide technical assistance. e. Personnel from other relevant departments and agencies, as necessary.

8. Requires the Secretary of GovOps to select eight members of the consortium, and the President pro Tempore of the Senate and the Speaker of the Assembly to each select three members.

9. Specifies that if CalCompute is established within the UC, the UC may receive

private donations for the purpose of implementing CalCompute.

SB 53 (Wiener)

Page 4 of 11

10. Provides that the above provisions shall become operative only upon an

appropriation in a budget act, or other measure, as specified.

11. Prohibits a developer, as defined, from making, adopting, or enforcing a rule, regulation, or policy that prevents an employee from disclosing, or from retaliating against an employee for disclosing, information to the Attorney General (AG), federal authorities, or another employee who has authority to investigate, discover, or correct the reported issue, as specified.

12. Authorizes an employee, as defined, to use the whistleblower hotline

maintained by the AG to make reports described above.

13. Requires a developer to provide a clear notice to all employees of their rights

and responsibilities under this bill, as specified.

14. Requires a developer to provide a reasonable internal process through which an

employee may anonymously disclose information to the developer if the employee believes in good faith that the information indicates that the developer’s activities present a critical risk, as specified.

15. Requires the disclosures and responses of the process required by this bill to be shared with officers and directors of the developer at least once each quarter, as specified.

16. Authorizes the court to award reasonable attorney’s fees to a plaintiff who

brings a successful action for a violation of this section.

17. Provides that in a civil action brought pursuant to this bill, the developer shall have the burden of proof to demonstrate by clear and convincing evidence that the alleged action would have occurred for legitimate, independent reasons even if the employee had not engaged in activities protected by this bill, as specified.

18. Provides that in a civil action or administrative proceeding brought pursuant to this bill, an employee may petition the superior court in any county wherein the violation in question is alleged to have occurred, or wherein the person resides or transacts business, for appropriate temporary or preliminary injunctive relief.

19. Provides, that upon the filing of the petition for injunctive relief, the petitioner

shall cause notice thereof to be served upon the person, and thereupon the court shall have jurisdiction to grant temporary injunctive relief as the court deems just and proper.

SB 53 (Wiener)

Page 5 of 11

20. Provides that in addition to any harm resulting directly from a violation of this bill, the court shall consider the chilling effect on other employees asserting their rights under this bill in determining whether temporary injunctive relief is just and proper.

21. Requires appropriate injunctive relief be issued on a showing that reasonable

cause exists to believe a violation has occurred.

22. Requires an order authorizing temporary injunctive relief to remain in effect

until an administrative or judicial determination or citation has been issued, or until the completion of a review, as specified, whichever is longer, or at a certain time set by the court. Thereafter, a preliminary or permanent injunction may be issued if it is shown to be just and proper. Any temporary injunctive relief shall not prohibit a developer from disciplining or terminating an employee for conduct that is unrelated to the claim of the retaliation.

23. Provides that injunctive relief granted pursuant to this bill shall not be stayed

pending appeal, as specified.

24. Includes relevant definitions of key words and terms, as specified.

Background

Author Statement. According to the author’s office, “the greatest innovations happen when our brightest minds have the resources they need and the freedom to speak their minds. California’s leadership on AI is more critical than ever as the new federal Administration proceeds with shredding the guardrails meant to keep Americans safe from the known and foreseeable risks that advanced AI systems present.”

Further, “Senate Bill 53 recognizes that we need a multipronged approach to encourage responsible AI innovation. It does so by: 1. Ensuring employees of frontier model labs are not retaliated against for reporting their belief that a developer’s activities pose a critical risk, as defined, or has made false or misleading statements about its management of critical risk; and, 2. Establishing a process to create a public cloud-computing cluster (CalCompute) that will conduct research into the safe and secure deployment of large-scale AI models.”

Cluster Computing. Cluster computing is a type of computing where multiple computers are connected to work together as a single system. The term “cluster” refers to the network of linked computer systems programmed performing the same task. Computing clusters typically consist of servers, workstations, and

SB 53 (Wiener)

Page 6 of 11

personal computers that communicate over a local area network or a wide area network. Cluster computing is a kind of distributed computing, a type of computing that links computers together on a network to perform a computational task, increase computational power and function as a single computer. Each computer, or “node,” in a cluster network has an operating system and a central processing unit core that handles the tasks required for the software to run properly.

Because of its high performance and high availability, cluster computing has many applications, including cloud computing, AI, scientific research, and big data analytics. Cloud computing is the on-demand access of computing resources – physical or virtual servers, data storage, networking capabilities, application development tools, software, AI-powered analytic platforms and more – over the internet with pay-per-use pricing. When you use cloud service, you’re accessing remote servers, powerful mainframe computers housed in large data centers, through the internet. The cloud computing model gives users greater flexibility and scalability compared to traditional on-premises infrastructure.

Examples include Amazon Web Services (AWS), which provides numerous clusters through services like Elastic Compute Cloud (EC2) and AWS Lambda. These cloud clusters power everything from simple web applications to complex AI training tasks. Google Cloud Platform (GCP) supports services like Google Compute Engine and Kubernetes Engine to simplify the deployment, management, and scaling of containerized applications, offering a scalable and automated environment for running containerized workloads. Microsoft Azure offers clusters through its Virtual Machines and Azure Kubernetes Service providing extensive AI and machine learning services integrated within its cloud environment. IBM Cloud and Oracle Cloud offer scalable computing clusters that support high- performance computing, big data applications, and AI workloads.

CalCompute. This bill requires GovOps to establish a consortium to develop a framework for the creation of a public cloud computing cluster to be known as “CalCompute.” By January 1, 2027, GovOps must submit to the Legislature the consortium’s report with the framework. This bill requires the framework for CalCompute to advance the development and deployment of AI that is safe, ethical, equitable, and sustainable by fostering research and innovation that benefits the public and enabling equitable innovation by expanding access to computational resources.

The consortium will be consist of eight members selected by the Secretary of GovOps, and three each by the President pro Tempore of the Senate and the Speaker of the Assembly. Specifically, this bill requires the 14 members be

SB 53 (Wiener)

Page 7 of 11

selected among representatives of the UC and other public and private academic research institutions and national laboratories; representatives of impacted workforce labor organizations; representatives of stakeholder groups with relevant expertise and experience; experts in technology and AI to provide technical assistance; and personnel from other relevant departments and agencies, as necessary.

This bill does not establish CalCompute, however, it does specify that if CalCompute is eventually established within the UC, the UC may receive private donations for the purposes of implementing CalCompute.

Supporters argue that a public CalCompute is a step towards democratizing access to advanced computational resources currently out of reach for many academics and startups in California. CalCompute can potentially lower the barriers to entry for innovative research and development in AI, and a shared resource can foster cross-sector collaboration, driving economic growth by leveling the playing field for smaller players that might otherwise be outpaced by large private firms.

Frontier Models and Safety. Frontier AI models are the cutting-edge class of “foundation” or general-purpose AI systems. These models have the ability to perform a wide variety of tasks – from natural language understanding to code generation and image creation – and are trained with incredible amounts of compute, data, energy, and water. These models are distinguished by their breadth of functionality and the scale of resources required to develop them (often costing hundreds of millions or billions of dollars). Their ability to produce powerful new capabilities potentially pose severe risks to safety or security if misused. Model derivatives are any downstream version of a frontier model created by modifying or adapting the original model’s weights, architecture, or outputs for specialized purposes and applications.

Frontier models, if misused, can give rise to serious harm such as launching large- scale cyberattacks that disrupt critical infrastructure and civil society – including power grids, communication networks, and financial systems – resulting in widespread economic and social chaos and uncertainty. These models can also be utilized to create autonomous systems capable of inflicting physical harm, for example, through the development of lethal autonomous weapons, the manipulation of transportation systems or dangerous recommendations from “therapy” chatbots. Additionally, and possibly most acutely, these models can be used to facilitate sophisticated disinformation campaigns that undermine public trust in democratic institutions destabilizing the political process. Other risks include privacy violations where sensitive or personal data might be exposed or misused.

SB 53 (Wiener)

Page 8 of 11

Examples of frontier models include OpenAi’s ChatGPT, Google’s Gemini, Anthropic’s Claude, Meta’s Llama, X-Ai’s Grok, and Hangzhou’s Deep Seek. These models represent the frontier in natural language processing, having been trained on vast datasets and massive computational resources, enabling them to perform a wide variety of autonomous tasks.

Critical Risk and AI. This bill strengthens whistleblower protections for employees of frontier AI laboratory companies whose activities developing frontier AI pose a critical risk, as defined. These protections apply to companies that have trained at least one foundation model with a quantity of computational power that costs at least $100 million. Specifically, this bill prevents developers of these frontier models from making, adopting, or enforcing any rule, regulation, or policy that prevents an employee from disclosing, or retaliates against an employee for disclosing, information to the AG, federal authorities, or another employee, as specified, if the employee has reasonable cause to believe that the information discloses that the developer’s activities pose a critical risk or the developer has made false or misleading statements about its management of critical risk.

Should this bill advance in this committee, it will be referred to the Senate Committee on the Judiciary which has appropriate jurisdiction to further discuss the whistleblower provisions of this bill.

Back to the Future. SB 1047 (Wiener, 2024) of last year would have, among other things, required developers of large AI models, and those providing the computing power to train such models, to put certain safeguards and policies in place to prevent catastrophic harm, as defined. Similarly to this bill, SB 1047 would have also established a consortium in GovOps to develop a framework for the creation of CalCompute and included enhanced whistleblower protections for employees that have reasonable cause to believe a developer is out of compliance with certain requirements for that covered AI model, as specified.

SB 1047 was vetoed by Governor Newsom who wrote, in part:

Proactive guardrails should be

Let me be clear – I agree with the author – we cannot afford to wait for a major catastrophe to occur before taking action to protect the public. California will not abandon its responsibility. Safety protocols must be adopted. implemented, and severe consequences for bad actors must be clear and enforceable. I do not agree, however, that to keep the public safe, we must settle for a solution that is not informed by an empirical trajectory analysis of AI systems and capabilities. Ultimately, any framework for effectively regulating AI needs to keep pace with the technology itself.

SB 53 (Wiener)

Page 9 of 11

To those who say there’s no problem here to solve, or that California does not have a role in regulating potential national security implications of this technology, I disagree. A California-only approach may well be warranted – especially absent federal action by Congress – but it must be based on empirical evidence and science. The U.S. AI Safety Institute, under the National Institute of Science and Technology, is developing guidance on national security risks, informed by evidence-based approaches, to guard against demonstrable risks to public safety. Under and Executive Order I issued in September 2023, agencies within my Administration are to performing risk analyses of potential California’s critical infrastructure using AI. These are just a few examples of the many endeavors underway, led by experts, to inform policymakers on AI risk management practices that are rooted in science and fact. And endeavors like these have led to the introduction of over a dozen bills regulating specific, known risks posed by AI, that I have signed in the last 30 days.

threats and vulnerabilities

Suggested Amendments. The author has agreed to the following amendments:

Amendment #1: 11547.61(h) The members of the consortium shall serve without compensation, but shall be reimbursed for all necessary expenses actually incurred in the performance of their duties.

Amendment #2: 11547.61(i) The working group shall be dissolved upon submission of the report required by paragraph (1) of subdivision (f) to the Legislature.

Prior/Related Legislation

SB 579 (Padilla, 2025) requires the Secretary of GovOps to appoint a mental health and AI working group to evaluate certain issues and determine the role of AI in mental health settings, as specified. (Pending in the Senate Governmental Organization Committee)

AB 1064 (Bauer-Kahan, 2025) the Leading Ethical AI Development (LEAD) for Kids Act, among other things, establishes the LEAD for Kids Standards Board, in GovOps, to adopt regulations governing criteria for determining the level of estimated risk of a covered product based on an analysis that weighs the likelihood and severity of reasonably foreseeable adverse impacts against the anticipated benefits of the covered product and denominating the risk levels, as specified. (Pending in the Assembly Privacy and Consumer Protection Committee)

SB 53 (Wiener)

Page 10 of 11

SB 893 (Padilla, 2023) would have required GovOps, the Governor’s Office of Business and Economic Development, the California Privacy Protection Agency, and CDT to establish the California AI Research Hub, as specified. (Held in the Assembly Appropriations Committee Suspense File)

SB 1047 (Wiener, 2024) would have required developers of powerful AI models, as defined, and those providing the computer power to train such models to put appropriate safeguards and polici4es in place to prevent critical harms, as defined. Would have established a state entity to oversee the development of these models and called for the creation of a consortium to develop a framework for a public cloud computing cluster. (Vetoed by Governor Newsom)

AB 2885 (Bauer-Kahan, Chapter 843, Statutes of 2024) defined “artificial intelligence” to mean an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments.

SCR 17 (Dodd, Res. Chapter 135, Statutes of 2023) affirms the California Legislature’s commitment to President Biden’s vision for a safe AI and the principles outlined in the “Blueprint for an AI Bill of Rights” and expresses the Legislature’s commitment to examining and implement those principles in its legislation and policies related to the use and deployment of automated systems.

FISCAL EFFECT: Appropriation: No Fiscal Com.: Yes Local: No

SUPPORT:

Secure AI Project (Co-source) Economic Security California Action (Co-source) Encode (Co-source) AI Policy Tracker Apart Research Center for AI and Digital Policy Center for Human-Compatible Artificial Intelligence, UC Berkeley Compute Exchange Nonlinear Oakland Privacy Redwood Research Safe AI Future

SB 53 (Wiener)

OPPOSITION:

None received

Page 11 of 11

ARGUMENTS IN SUPPORT: In support of the bill, the co-sources write that, “AI researchers and engineers will likely be the first to know if a company’s product poses a significant risk to public safety, but company contracts may prevent them from reporting concerns. In general, companies offer or require broad non-disclosure and non-disparagement agreements that are quite broad, often limiting nearly all disclosures that are not explicitly protected by law. These agreements in practice can prevent employees from sharing a good faith public safety concern with a government authority, which could chill employees’ disclosure rights when that is their best or only course of action.”

In addition to “whistleblower protections, SB 53 would also set in motion CalCompute, a public cloud computing cluster for use by academics and startups in California. Computational resources are essential for AI research, and CalCompute would make those resources more accessible to California’s top universities and startups, helping to catalyze additional research into beneficial applications of AI and supporting in particular smaller startups for a healthier innovation ecosystem. We support this groundbreaking effort, which would advance and democratize AI research in California.”

DUAL-REFERRAL: Senate Committee on Governmental Organization & Senate Committee on Judiciary