AI Laws in San Francisco, California
AI Laws USA tracks 92 AI rules that apply to people and businesses in San Francisco, California: 10+ federal rules, 73 California state-level rules, and 9 local San Francisco ordinances. Coverage is strongest on consumer protection, automated decision-making, AI disclosure and transparency, and consumer data privacy. 54 of these rules are already in effect. Each entry below links to its official source.
San Francisco local AI rules (and San Francisco County County)
9 local AI rules specific to San Francisco, California or San Francisco County County.
-
Enacted (not yet in effect) Strength: moderate
San Francisco, CA — 45-day data center moratorium (approved Oct. 6, 2026)
San Francisco, CA · Enacted October 6, 2026 · San Francisco, Cal. — proposed 45-day interim data center moratorium (introduced Sept. 15, 2026)
San Francisco's Board of Supervisors unanimously approved a 45-day moratorium on approving new data centers on Oct. 6, 2026. Supervisor Shamann Walton's measure covers data centers of any size, and the Planning Department must report back within 25 days on defining data centers and studying their energy, water, air quality and community effects. The SF Standard reported it takes effect within 10 days unless the mayor returns it sooner, and it can be extended for up to about two years.
-
Enacted (not yet in effect) Strength: limited
San Francisco ALPR privacy protections (Sept 2026)
San Francisco, CA · Enacted September 29, 2026 · Office of the Mayor, City and County of San Francisco, press release on public safety camera privacy protections (Sept. 29, 2026)
On September 29, 2026, Mayor Daniel Lurie announced tighter rules for how San Francisco police use license plate reader data. Officers can now search plate-reader data for investigations only during the first 30 days after it is captured, down from 12 months, and each search must be tied to a dispatch or incident number. SFPD's data is to be removed from the vendor's platform after 30 days, federal agencies get no direct access, and use for civil immigration enforcement, tracking reproductive or gender-affirming care, or monitoring First Amendment activity is prohibited. SFPD will review system-use logs monthly and will ask the Department of Police Accountability to conduct regular independent audits; personnel who violate the rules face discipline up to termination. The Electronic Frontier Foundation criticized the package as weak because it does not require a warrant or set a deletion deadline.
-
Proposed / pending
SFPD proposed DGO 10.12 drone policy (2026)
San Francisco, CA · San Francisco Police Department, proposed Department General Order 10.12, Unmanned Aircraft Systems (UAS) Operations (Police Commission, Sept. 2, 2026; hearing set Oct. 14, 2026)
The San Francisco Police Department has proposed a new general order, DGO 10.12, setting rules for its drones. The draft would let officers use drones in any situation where an officer may be sent for a public safety response, while banning uses such as monitoring protests solely for that purpose, targeting protected classes, use as a force option and 'general surveillance'. Footage with no evidentiary value would be kept no more than 30 days and evidence footage at least two years. The Police Commission delayed a vote on September 2, 2026 to take more public comment and is scheduled to hear it again on October 14, 2026.
-
In effect Strength: stronger
SF Stop Secret Surveillance Ordinance (2019, first-in-nation)
San Francisco, CA · Effective July 15, 2019 · S.F. Ordinance No. 107-19 (2019); S.F. Admin. Code ch. 19B
San Francisco became the first city in the United States to ban its own government — including police — from using facial recognition technology. The Board of Supervisors gave final passage 10-1 on June 4, 2019; the ordinance became law without the mayor's signature on June 14, 2019 and took effect July 15, 2019, codified at SF Admin. Code Chapter 19B. It also requires Board approval and public use policies before city departments acquire any other surveillance technology. Still in effect 2026 and the template for municipal facial-recognition bans nationwide.
-
In effect Strength: limited
SF Generative AI Guidelines
San Francisco, CA · Effective December 11, 2023 · CCSF, Generative AI Guidelines (Dec. 2023, rev. July 2025)
San Francisco's citywide generative AI guidelines apply to employees, contractors, consultants, volunteers, and vendors working for the city. They require human review and disclosure of AI-generated content, prohibit entering non-public information into AI tools, ban concealing AI use, and bar generating deepfake-style images, audio, or video. Most recently revised in July 2025.
-
In effect Strength: limited
San Francisco Unified School District
San Francisco, CA · Effective November 12, 2024 · San Francisco Unified School District — Generative AI Use Guidelines (2024-11-12)
District guidelines authorize Microsoft Copilot enterprise for staff; bar student AI accounts under 13; require teacher disclosure when AI is used for instructional design; bar AI use to make discipline or placement decisions without human review.
-
In effect Strength: moderate
SF delivery-robot permit (2017)
San Francisco · Effective January 21, 2018 · S.F. Pub. Works Code § 794
San Francisco became the first U.S. city to comprehensively restrict sidewalk delivery robots after constituent backlash. Supervisor Norman Yee's ordinance limits autonomous delivery devices to a small permit program (initially 9 city-wide and 3 per company), bans them from most sidewalks, caps speed at 3 mph in pedestrian zones, and requires a human chaperone.
-
In effect Strength: limited
Ordinance prohibiting sale or use of algorithmic devices to
San Francisco, CA · Effective October 14, 2024 · Ordinance prohibiting sale or use of algorithmic devices to set rents (Administrative Code / Rent Ordinance Section 37.10C)
San Francisco, the first US city to do so, bans landlords from selling or using algorithmic revenue-management software that uses non-public competitor data to recommend rents or occupancy levels, with civil penalties up to $1,000 per violation plus damages and attorneys' fees.
-
In effect Strength: limited
SFPD Unmanned Aircraft System (Drone) Program authorized und
San Francisco, CA · Effective May 16, 2024 · SFPD Unmanned Aircraft System (Drone) Program authorized under Proposition E (SF Admin Code 96I.2)
After voters passed Proposition E in March 2024, SFPD began operating drones for criminal investigations, vehicle pursuits, and critical incidents, with publicly available flight logs starting May 16, 2024.
California-level AI rules most relevant to San Francisco
73 California state rules apply to residents and businesses in San Francisco. Showing the 8 most relevant to San Francisco's local picture; 65 more are on the California jurisdiction page.
-
In effect Strength: limited
CCC systemwide AI guidance and HUMANS responsible-AI framewo
California Community Colleges Chancellor's Office · Effective January 1, 2024 · CCC systemwide AI guidance and HUMANS responsible-AI framework
The California Community Colleges Chancellor's Office issues systemwide AI guidance built on its HUMANS framework (human-centered, privacy, algorithmic-discrimination protections, notice and explanation, safety) governing AI in instruction and student support.
-
In effect Strength: limited
Executive Order N-5-26 - Trusted AI Procurement
California · Effective March 30, 2026 · Executive Order N-5-26 - Trusted AI Procurement
This newer California executive order directs DGS and CDT to develop trust-and-safety certifications for state AI contracting (covering CSAM/NCII, harmful bias, and civil-rights violations), reforms to bar contracting with entities that unlawfully undermine privacy or civil liberties, and CDT guidance on watermarking AI-generated media.
-
In effect Strength: limited
UC Responsible AI Principles and AI Council governance frame
University of California (UC System) · Effective October 6, 2021 · UC Responsible AI Principles and AI Council governance framework
UC adopted its Responsible AI Principles on October 6, 2021 (describing itself as one of the first universities to do so) and established a systemwide AI Council in May 2022 that issues guidance, training, and risk assessments for AI use across its campuses.
-
In effect Strength: limited
BART Surveillance Technology Ordinance
Bay Area Rapid Transit (BART) · Effective September 13, 2018 · BART Surveillance Technology Ordinance
BART became the first transit district in the country to adopt a CCOPS-style ordinance requiring board approval, a surveillance impact report, a use policy, and annual reports before acquiring surveillance technology.
-
In effect Strength: limited
Attorney General Bonta Legal Advisories on the Application o
California · Effective January 13, 2025 · Attorney General Bonta Legal Advisories on the Application of California Law to AI
California Attorney General Rob Bonta issued two legal advisories clarifying that entities developing, selling, or using AI must comply with existing California consumer-protection, civil-rights, competition, data-privacy, and election-misinformation laws, plus new AI laws effective January 1, 2025, with a second advisory targeting healthcare entities.
-
In effect Strength: limited
Guidance for the Safe and Effective Use of Artificial Intell
California Department of Education · Effective January 1, 2026 · Guidance for the Safe and Effective Use of Artificial Intelligence in California Public Schools
California's education department issued voluntary guidance for K-12 districts covering human-centered AI, AI literacy, equitable access, academic integrity, data privacy (FERPA/COPPA/CCPA/SOPIPA), and procurement.
-
In effect Strength: stronger
CCPA/CPRA + ADMT Regulations
California · Effective January 1, 2026 · Cal. Civ. Code § 1798.100 et seq.; Cal. Code Regs. tit. 11, div. 6
California's main privacy law gives consumers rights to know, delete, correct, and opt out of the sale or sharing of their personal information. New regulations finalized in 2025 add rights around automated decision-making technology (ADMT): businesses using ADMT for significant decisions (jobs, housing, credit, healthcare) must give pre-use notice, let people opt out, and provide access to how decisions were made.
-
In effect Strength: limited
California GenAI Procurement Guidelines and Toolkit
California · Effective March 21, 2024 · California GenAI Procurement Guidelines and Toolkit
California's Government Operations Agency and Department of Technology issued procurement guidelines and a toolkit requiring state entities to complete a SIMM 5305-F GenAI risk assessment, use written solicitations with a GenAI Disclosure & Fact Sheet, engage the CIO/AIO, and report GenAI contracts before procuring generative AI.
Federal AI rules that apply in San Francisco, California
These federal rules apply everywhere in the United States, including San Francisco, California. Showing the 10 strongest and most recent.
-
In effect Strength: stronger
TVA data center capacity commitment charge (2026)
Tennessee Valley Authority · Effective October 1, 2026 · Tenn. Valley Auth. Bd. of Dirs. — Data Center Capacity Commitment Charge (approved Aug. 20, 2026; eff. Oct. 1, 2026)
The Tennessee Valley Authority's Board of Directors approved a separate, higher wholesale rate for data centers on August 20, 2026, effective October 1. New or expanding data center loads above 5 megawatts now fall under a Capacity Commitment Charge attached to TVA's data center rate schedules, designed to recover the incremental capacity costs those customers impose on the system. The headline figures are substantial: upfront capacity commitment charges of roughly $1.5 million per megawatt, payable over three to five years, and an all-in average billing increase for data center customers of about 10%, phased in across three consecutive fiscal years. The stated purpose is to keep residential rates low while still serving growing data center demand. TVA matters here out of proportion to a single utility. It is a federal corporate agency supplying power across seven states, so this is not a state rate case but a regional cost-allocation decision reaching a large share of the Southeast — and TVA is among the first large US utilities to adopt a tariff aimed specifically at AI-driven load. Where most 2026 measures in this dataset restrict where data centers may be built, this one changes what they pay to run.
-
In effect Strength: stronger
Bartz v. Anthropic
N.D. Cal. · Effective September 5, 2025 · Bartz v. Anthropic PBC, No. 3:24-cv-05417 (N.D. Cal.)
Authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson sued Anthropic over its use of pirated-book datasets to train Claude. In June 2025 Judge William Alsup ruled that training on lawfully purchased books was fair use. The piracy claims (LibGen ingestion) proceeded toward settlement; in September 2025 Anthropic agreed to a $1.5 billion class settlement covering ~500,000 works. Judge Alsup granted preliminary approval September 25, 2025 and final approval July 20, 2026. The appeal period runs through August 19, 2026; class members will not receive payments until at least end of 2026.
-
In effect Strength: stronger
COPPA + 2025 Rule (childrens data)
United States · Effective June 23, 2025 · 15 U.S.C. §§ 6501–6506; 16 C.F.R. Part 312
COPPA requires online services aimed at children under 13 to get verifiable parental consent before collecting kids' personal data. The 2025 rule update — fully in effect since April 22, 2026 — adds biometric identifiers (like face templates and voiceprints, which matter for AI tools), requires separate parental consent before sharing children's data for targeted advertising, and tightens data retention limits.
-
In effect Strength: stronger
TAKE IT DOWN Act
United States · Effective May 19, 2025 · Pub. L. No. 119-12 (S. 146)
Makes it a federal crime to knowingly publish intimate images of someone without consent, including AI-generated deepfakes. Social media and similar platforms must give victims a way to request removal and must take the content (and known copies) down within 48 hours. The platform removal requirement became enforceable May 19, 2026, and the FTC has already begun enforcement.
-
Blocked / in litigation
NetChoice v. Yost (Ohio)
S.D. Ohio · Effective April 16, 2025 · NetChoice, LLC v. Yost, No. 2:24-cv-00047 (S.D. Ohio Apr. 16, 2025)
Ohio's Social Media Parental Notification Act — requiring parental consent for minors' social-media use, including algorithmic feeds — was preliminarily enjoined on February 12, 2024, then permanently enjoined on April 16, 2025 when the district court granted summary judgment for NetChoice. The state appealed to the Sixth Circuit, which vacated the district court's injunction in 2026.
-
In effect Strength: stronger
Thaler v. Perlmutter (Copyright)
D.C. Cir. · Effective March 18, 2025 · Thaler v. Perlmutter, 130 F.4th 1039 (D.C. Cir. 2025)
The companion copyright case: Stephen Thaler sought to register a copyright with 'Creativity Machine' (his AI) as the author. The D.C. Circuit affirmed in March 2025 that the Copyright Act's human-authorship requirement is dispositive as a matter of statutory law. AI cannot be a copyright author under U.S. law.
-
In effect Strength: stronger
Louis v. SafeRent
D. Mass. · Effective November 20, 2024 · Louis v. SafeRent Solutions, LLC, No. 1:22-cv-10800 (D. Mass.)
SafeRent agreed in November 2024 to a $2.275M settlement and a five-year ban on using its 'SafeRent Score' for housing-voucher applicants, after a class action alleged its AI tenant-screening tool systematically denied housing to Black and Hispanic Section 8 voucher holders. The first major AI tenant-screening Fair Housing Act settlement.
-
In effect Strength: stronger
FTC Impersonation Rule (AI)
United States · Effective April 1, 2024 · 16 C.F.R. Part 461; 89 Fed. Reg. 15017
The FTC's Impersonation Rule lets the agency directly sue scammers who pretend to be a government agency or a real business — including those who use AI-cloned voices or generated images to do so. Civil penalties can reach $53,088 per violation. The FTC also issued a supplemental notice in February 2024 proposing to extend the rule to all individual impersonation.
-
In effect Strength: stronger
FTC v. Rite Aid
FTC · Effective February 26, 2024 · FTC v. Rite Aid Corp., No. 2:23-cv-05023 (E.D. Pa. Dec. 19, 2023)
The FTC banned Rite Aid from using facial-recognition technology in its stores for five years after finding the pharmacy chain's FRT system falsely tagged customers — disproportionately women and people of color — as shoplifters, leading to wrongful detentions and humiliations.
-
In effect Strength: stronger
TCPA (AI voice calls)
United States · Effective February 8, 2024 · 47 U.S.C. § 227; FCC 24-17
Robocalls using AI-cloned or AI-generated voices are treated like other 'artificial voice' calls: callers need your prior express consent, must identify themselves, and must offer opt-outs for telemarketing. You can personally sue violators for $500 to $1,500 per illegal call.
Frequently asked questions about AI laws in San Francisco, California
Are there AI laws in San Francisco, California?
What federal AI rules apply in San Francisco?
Does California have an AI privacy law?
Are deepfakes illegal in California?
Can my employer use AI to screen me for jobs in San Francisco?
How do I report an AI law violation in San Francisco?
Are facial recognition cameras allowed in San Francisco?
Is San Francisco regulated by California's consumer privacy act?
Compliance guides for California businesses
Plain-English guides to the California AI and data laws companies most often need to comply with:
California ADMT / automated-decision rules · California deepfake law for business
Have we missed an AI rule in San Francisco?
This page is generated from our open civic dataset. If you know of a San Francisco ordinance, county rule, or local enforcement action we should add, email feedback@ailawsusa.com or submit a correction. Every entry must include a verifiable source.