Skip to main content

Local AI hub

AI in San Francisco, California

San Francisco, CA is a major regional employment centre, with San Francisco County home to 836,321 residents at the 2023 American Community Survey. Median home value near $1,380,500 puts algorithmic valuation, tenant screening and lending models directly in the path of ordinary transactions here. Median household income above $141,446 means people affected by an automated decision are more likely to contest it, which raises the standard of explanation a system has to meet.

San Francisco, California by the numbers

US Census Bureau, American Community Survey 2023, San Francisco County.

Population
836,321
Median household income
$141,446
Unemployment rate
5.6%
Largest industry
Professional/Management

Read the full AI ethics guide for San Francisco County

Government

  • San Francisco Generative AI Guidelines (2025 revision)

    San Francisco issued the first citywide generative AI guidelines in the United States in 2023 and revised them in July 2025. Two prohibitions are stated outright: staff may not rely on AI to produce official documents or make decisions without expert human review, and may not generate images, audio or video that could be mistaken for real people, public officials included. Naming synthetic media of officials specifically is rare, and it is the clause other cities are most likely to need.

    Source: City and County of San Francisco · verified August 2026

  • San Francisco 22J AI inventory process

    Public-facing or sensitive uses of generative AI — drafting for hiring, enforcement or public communication — must be documented through the city 22J inventory process, while low-risk internal work such as memo drafting or summarising needs no formal disclosure provided the output is verified. Splitting the duty by consequence rather than by tool is the design most municipal policies get wrong, and this is a working example of getting it right.

    Source: City and County of San Francisco · verified August 2026

  • San Francisco Generative AI Guidelines, original 2023 edition

    The December 2023 edition, revised from a 9 November 2023 draft, remains worth reading beside the current one. Comparing the two shows which parts of a first municipal AI policy survived eighteen months of actual use and which did not — evidence almost no other city has published, because almost no other city has a second edition.

    Source: City and County of San Francisco · verified August 2026

  • SFPD Department Bulletin 24-009 — generative AI guidelines for staff

    The Police Department issued its own bulletin implementing the citywide guidelines. Departmental adoption is where municipal AI policy usually stops being real, and a published law-enforcement bulletin is the strongest evidence available that San Francisco guidance reached the departments where the stakes are highest.

    Source: San Francisco Police Department · verified August 2026

  • California Executive Order N-5-26 — trusted AI procurement

    Signed 30 March 2026, this order requires any company selling AI to the State of California to certify that its systems carry safeguards against illegal content, harmful bias, and violations of civil rights and liberties. For an organisation in San Francisco County the practical reach is wider than state agencies: procurement certification language is routinely inherited by counties and cities writing their own contracts, so this is the wording local buyers are most likely to copy.

    Source: Office of the Governor of California · verified August 2026

  • California SAM 4986 — state GenAI policy suite

    The Department of Technology sets the operative rules for generative AI in California government through the State Administrative Manual, effective 20 February 2025. SAM 4986.1 governs responsible and secure use; SAM 4986.4 requires transparency about training data, with documentation posted publicly from 1 January 2026. Public bodies in San Francisco County sit downstream of both, and private employers get a published, free drafting reference that state oversight already recognises.

    Source: California Department of Technology · verified August 2026

  • California AB 2013 — generative AI training data disclosure

    AB 2013 requires public disclosure of the data used to train generative AI systems, including summaries of the datasets and their sources. It is one of the few obligations that reaches model developers rather than deployers, which matters in San Francisco County because it changes what a buyer there can reasonably demand: a supplier subject to AB 2013 has already had to produce the disclosure, so asking for it is asking for something that exists.

    Source: California Legislature / CDT · verified August 2026

  • California Executive Order N-12-23 — the GenAI framework

    The September 2023 order that established how California prepares for generative AI, and the instrument the later procurement and workforce orders build on. Reading it first makes the rest legible. For San Francisco County organisations it is the clearest statement of the state posture: adopt deliberately, assess risk before deployment, and treat public-sector use as requiring justification rather than permission.

    Source: California Government Operations Agency · verified August 2026

Training

  • Enterprise generative AI tooling for San Francisco city staff

    The Department of Technology procures and licenses enterprise generative AI tools for city staff, so use happens inside a governed environment rather than through personal accounts. Providing a sanctioned tool is what makes a prohibition enforceable; a policy that forbids consumer tools without supplying an alternative mostly produces unlogged use.

    Source: City and County of San Francisco Department of Technology · verified August 2026

  • California AI workforce preparation executive order

    Signed 21 May 2026, this order directs the state to prepare workers, small businesses and communities for economic disruption from AI. It is the instrument that funds and frames retraining, so for employers in San Francisco County planning workforce change it identifies which state programmes are being pointed at the problem rather than leaving the question to guesswork.

    Source: Office of the Governor of California · verified August 2026

  • California judicial branch generative AI rules

    The California Courts published their own generative AI requirements, which matter to San Francisco County beyond the courts themselves: they are the clearest published example in the state of an institution that must accept AI-assisted work from outside parties setting out what it will and will not tolerate. Any organisation filing, reporting or contracting into a regulated process can read it as a floor.

    Source: Judicial Council of California · verified August 2026

Organization

  • AI@UCSF

    The University of California, San Francisco coordinates AI work across the Bakar Computational Health Sciences Institute, the Center for Intelligent Imaging and a joint computational precision health programme with UC Berkeley. In a city whose AI reputation is built on private labs, UCSF is the largest concentration of AI expertise locally that is not attached to a company selling something.

    Source: University of California, San Francisco · verified August 2026

  • Bakar Computational Health Sciences Institute, UCSF

    A UCSF institute applying computational and AI methods to health. For San Francisco organisations outside healthcare it still matters: clinical AI carries the strictest evidence expectations of any deployment domain, so the standards developed here tend to be the ceiling other sectors are eventually measured against.

    Source: University of California, San Francisco · verified August 2026

  • UCSF Center for Intelligent Imaging

    A UCSF centre on AI in medical imaging, one of the few AI application areas with a long record of regulatory scrutiny. It is a useful local reference for any San Francisco organisation trying to work out what "validated" ought to mean for a model, since imaging had to answer that question before most fields asked it.

    Source: University of California, San Francisco · verified August 2026

Other CA markets

About this hub

San Francisco, California is a local hub of AI Coalition Network, an independent research organisation that ranks AI companies on ethics and publishes plain-language guides to AI governance and policy. Local statistics come from the US Census Bureau and carry the survey year. Nothing here is legal, tax, accounting or compliance advice.

Editorial policy Report an error Machine-readable record