26-110 Department Artificial Intelligence Guidance

The purpose of this Department Notice is to inform employees that the SFPD Crime Strategies Division will be introducing a new Artificial Intelligence (AI) tool, which requires employees to be trained before receiving access. This notice also establishes department-approved guidelines governing the use of Generative Artificial Intelligence (GenAI) by SFPD employees. These guidelines are intended to ensure that employees use AI tools responsibly, lawfully, and in a manner that maintains public trust, safeguards sensitive information, and protects the integrity of City systems.

Definitions and Terms 

• Artificial Intelligence or “AI” is defined in Penal Code 311 and CA Govt. Code 11546.45.5. as an engineered or machine-based system that varies in its level of autonomy and can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments.
• Generative AI (GenAI) means the class of AI models that emulate the structure and characteristics of input data to generate derived synthetic content. This can include images, videos, audio, text, and other digital content.
• Agentic AI refers to AI systems designed to act autonomously, perceiving their environment, making decisions, and taking actions to achieve specific goals. These systems often incorporate features like adaptability, goal orientation, and interaction with dynamic environments.
• Prohibited Uses are tasks that are not approved to use GenAI tools to protect public trust, safety, and ethical standards.
• Passive AI Use is when AI operates in the background of software (i.e. automated summaries, tagging, or prioritization).
• Qualified member is an individual who is deemed competent and authorized to perform specific functions based on pre-defined requirements such as skills, experience, profession or licensing.

Overview

SFPD recognizes the potential for the use of AI in various aspects of work within the Department. Employees must thoroughly review all AI outputs and ensure their work complies with Department policies, confidentiality rules, and all applicable City requirements outlined in the San Francisco Generative AI Guidelines including the AI Transparency Ordinance (Chapter 22J), which mandates Departmental disclosure when AI is used in work that affects the public. For more guidance and information, please read the full Generative AI Guidelines. Employees are reminded that the use of any tool—GenAI or otherwise—does not alter their responsibility to exercise sound judgement and comply with Department directives.

Guidelines

The City and County of San Francisco has issued guidelines to all city departments relating to the use of AI. Following these guidelines, all employees shall:

1. Always review, fact-check, and verify AI-generated information before relying on it to support decisions, especially in high-risk use cases and when using Agentic AI.
2. Adhere to City guidance on AI disclosure. Passive AI use disclosure will be implemented through system labeling, metadata, or audit logs.
3. Not use AI to create any work product that must be attested to under penalty of perjury, including warrants and incident reports. The Chief of Police may authorize a temporary suspension of this requirement for decision-making purposes while piloting or testing new technology.
4. Never enter sensitive information into unauthorized Generative AI tools. The information entered can be viewed by the companies that make the tools and, in some cases, members of the public. Employees are also reminded that CLETS information shall not be loaded into unauthorized systems at any time, per DN 25-037 CLETS - Protecting Sensitive Criminal Justice Information.

High-Risk Uses

Employees must exercise extra caution when using SFPD-specific GenAI tools for tasks that influence operational decisions and are made quickly under time constraints. These supplements do not replace judgement. Some examples include:

• Supporting decisions related to services, enforcement, or eligibility.
• Using AI-generated historical information.
• Supporting decisions during active operations.

All critical decisions must be made by a qualified member.

Prohibited Uses

To protect public trust, safety, and ethical standards, employees shall not use GenAI tools for the following:

• Creating official documents, decisions, or records without human review and approval.
• Generating images, audio, or video that could be mistaken for real individuals (including officers, City officials, or members of the public).
• Creating impersonations, or enhanced media for any purpose, including training or demonstrations.
• Fabricating public input, survey responses, or community feedback.
• Using AI to interpret legal, regulatory, or statutory requirements or to create legal conclusions.
• Generate official police reports, either fully or partially

Employees should be aware that on January 1, 2026, Senate Bill 524 enacted Penal Code § 13663 which applies when AI is used to generate reports, either fully or partially. The statute requires prominent disclosure of AI use, the ability to retain the initial AI-generated draft, and maintenance of an audit trail identifying the user and source material for the duration of the report’s retention period. Currently, the Department does not have a means to meet the statutory requirements of the law. As a result, Department employees shall not use artificial intelligence to generate official police reports, either fully or partially.

Restrictions on Public or Consumer AI Tools

Employees shall not enter sensitive or protected data that cannot be fully released to the public, including personal information, health information, City data, and/or financial information into public, free, or consumer AI platforms (including unapproved websites, mobile apps, or chatbots). These tools lack the required privacy protection, and information provided to them may be stored externally or used to train commercial models.

If used in accordance with this restriction, SFPD employees shall not conceal the use of public or consumer GenAI tools from coworkers and shall remain transparent about when and how they use these tools in their work.

For any questions regarding AI technology and/or to obtain a list of AI technology currently in use by SFPD (San Francisco AI Use Inventory (Chapter 22J) | DataSF), please contact [Redacted].

Administration

Supervisors shall ensure that employees understand and adhere to this notice. Misuse of AI tools, failure to safeguard data, or violation of City AI policy may result in administrative action.

/s/
DERRICK J. LEW
Chief of Police