San Francisco Generative AI Guidelines

• Authority: San Francisco, CA
• Parent authority: Local governments
• Jurisdiction: United States
• Collections: U.S. state and local documents
• Most recent activity: Proposed
• Most recent activity date: 2023-11-09

Risk subdomain coverage

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

• Compromise of privacy
• False or misleading information
• Lack of capability or robustness
• Lack of transparency or interpretability

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

• Unfair discrimination and misrepresentation
• Unequal performance across groups
• AI system security vulnerabilities
• Fraud, scams, and targeted manipulation
• Overreliance and unsafe use

All other risk subdomains had no coverage in this document.

Risk Coverage Summary

The document has good coverage of approximately 8-9 subdomains, with strong focus on misinformation (3.1), privacy compromise (2.1), security vulnerabilities (2.2), unfair discrimination (1.1), lack of robustness (7.3), and lack of transparency (7.4). Coverage is concentrated in misinformation, privacy/security, discrimination, and AI system reliability domains.

Sector coverage

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

• Public Administration (excl. National Security)

All other sectors had no coverage in this document.

Sectors Summary

This is an internal government policy document that governs AI use within San Francisco city government operations. The primary sector governed is Public Administration excluding National Security, as it applies to all city department personnel and their use of Generative AI in delivering public services.

Legislative Status

Other

Legislative Status Summary

This is an internal government policy document providing guidelines for city employees' use of Generative AI. It uses primarily mandatory language ('must', 'never', 'always') but lacks formal enforcement mechanisms, penalties, or sanctions typical of hard law.

Actors

Proposer

• Entity Types: AI_Governance_Actor
• Entities: City Administrator's Office; Digital & Data Services; Department of Technology; Committee on Information Technology
• Summary: The document explicitly identifies the City Administrator's Office (led by Digital & Data Services, the Department of Technology, and the Committee on Information Technology) as having developed these guidelines.

Target

• Entity Types: AI_User, AI_Deployer
• Entities: San Francisco city department personnel including employees, contractors, consultants, volunteers, and vendors
• Summary: The guidelines explicitly apply to all city department personnel who may use Generative AI tools in their work for the City and County of San Francisco.

Enforcer

• Entity Types: AI_Governance_Actor
• Entities: City Administrator's Office; Committee on Information Technology; Departmental IT Leaders; Office of Cybersecurity
• Summary: The City Administrator's Office oversees the guidelines and will update them. Departmental IT leaders have responsibility for supporting appropriate use cases. The Office of Cybersecurity is consulted for security matters.

Monitor

• Entity Types: AI_Governance_Actor
• Entities: City Administrator's Office; Committee on Information Technology; Departmental IT Leaders
• Summary: The document establishes monitoring responsibilities including collecting use cases, tracking usage, and reporting publicly. Departmental IT leaders are expected to track and report uses.

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 primarily addresses the Deploy and Operate and Monitor stages of the AI lifecycle, with some coverage of Verify and Validate. It focuses on the use of existing public Generative AI tools by city staff rather than the development of AI systems from scratch.

Technical Scope

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

• AI Models
• AI Systems
• Generative AI
• Predictive AI

Technical Scope Summary

The document explicitly focuses on Generative AI, defining it and distinguishing it from traditional discriminative AI. It mentions AI models, AI systems, and large language models (LLMs). It does not explicitly mention frontier AI, general purpose AI, task-specific AI, foundation models, or compute thresholds. It does not explicitly address open-weight or open-source models.