California SB 53

• Authority: California
• Parent authority: State governments
• Jurisdiction: United States
• Collections: Editors' Picks;U.S. state and local documents
• Most recent activity: Proposed
• Most recent activity date: 2025-01-07

Risk subdomain coverage

Good coverage — Comprehensive governance measures or mitigations with clear procedures relating to the risk subdomain are explicitly described.

• AI system security vulnerabilities
• Cyberattacks, weapon development or use
• Governance failure
• AI possessing dangerous capabilities

Minimal coverage — Brief mention but no specific mitigations, recommendations or governance measures relating to the risk subdomain are described.

• False or misleading information
• Power centralization and unfair distribution
• Environmental harm
• AI pursuing its own goals
• Lack of capability or robustness
• Lack of transparency or interpretability

All other risk subdomains had no coverage in this document.

Risk Coverage Summary

The document has good coverage of approximately 5-6 subdomains, with strong focus on malicious actors (4.2), AI system security (2.2), governance failure (6.5), dangerous capabilities (7.2), and lack of robustness (7.3). Coverage is concentrated in security, misuse prevention, and AI safety domains, particularly through whistleblower protections for critical risks.

Sector coverage

Good coverage — Comprehensive governance measures or mitigations with clear procedures relating to the sector are explicitly described.

• Information
• Scientific Research & Development Services

Minimal coverage — Brief mention but no specific mitigations, recommendations or governance measures relating to the sector are described.

• Educational Services
• Public Administration (excl. National Security)

All other sectors had no coverage in this document.

Sectors Summary

The document primarily governs the Information sector (AI developers and cloud computing infrastructure) and Scientific Research and Development Services (through CalCompute's research mission). It also has minimal coverage of Educational Services and Public Administration through CalCompute's establishment within the University of California and government oversight.

Legislative Status

Hard Law

Legislative Status Summary

This is a California state bill (SB 53) that establishes binding legal obligations on AI developers with mandatory language, enforcement mechanisms, and legal remedies including injunctive relief and attorney's fees.

Actors

Proposer

• Entity Types: AI_Governance_Actor
• Summary: The document is a California state bill (SB 53) which indicates it was proposed by the California Legislature. The consortium is established to develop the CalCompute framework with members selected by government officials.

Target

• Entity Types: AI_Developer
• Summary: The document explicitly defines and targets 'developers' of foundation models with specific computational thresholds, imposing whistleblower protection requirements and prohibiting retaliation. CalCompute is to be established within UC.

Enforcer

• Entity Types: AI_Governance_Actor
• Summary: The document establishes court jurisdiction for enforcement, references the Attorney General and Labor Commissioner as recipients of disclosures, and provides for civil actions and injunctive relief through the superior court system.

Monitor

• Entity Types: AI_Governance_Actor
• Summary: The Government Operations Agency must submit reports to the Legislature on CalCompute framework development. Internal monitoring is required through employee disclosure processes, with external oversight by the Attorney General.

AI Lifecycle Stages

Lifecycle stages covered by the documents, as defined in the OECD Framework for the Classification of AI Systems

• Plan and Design
• Collect and Process Data
• Build and Use Model
• Verify and Validate
• Deploy
• Operate and Monitor

AI Lifecycle Summary

The document covers multiple AI lifecycle stages with primary focus on Build and Use Model (through computational cost thresholds for foundation models), Deploy (through critical risk definitions related to deployment), and Operate and Monitor (through whistleblower protections for ongoing risk management and internal disclosure processes). CalCompute provisions address Plan and Design for future AI infrastructure.

Technical Scope

Terms relating to technical scope that are mentioned or implicitly referenced in the document

• AI Models
• Foundation Models
• AI Systems
• Compute Threshold

Technical Scope Summary

The document explicitly defines and covers AI models, foundation models, and establishes a computational cost threshold of $100,000,000 for training. It does not explicitly mention frontier AI, general purpose AI, task-specific AI, generative AI, predictive AI, or open-weight models. AI systems are mentioned in the context of CalCompute infrastructure.