California SB 942 Scope Checker

How to use

1. Question 1: say whether you produce a generative AI system that is publicly accessible in California and creates or alters images, video or audio (text-only systems follow a different path).
2. Questions 2–3: pin down your role — producer, as-is embedder, or white-label/fine-tuner — and whether the system is primarily assistive technology.
3. Questions 4–5: self-check your latent-disclosure fields and verification tool, and flag any platform, hosting or device-maker roles; then read your result tier, the gap list, and the key-dates table below.

About this tool

For a covered provider, two duties do the work. The latent disclosure (§ 22757.3(b)) embeds hidden, machine-readable provenance in generated or altered image, video and audio content: the provider’s name, the system’s name and version, the time and date, a unique identifier, and whether the system created or altered the content — permanent or extraordinarily difficult to remove, and interoperable with widely recognized standards such as C2PA Content Credentials. The disclosure verification tool (§ 22757.2) is a free public tool — upload, URL or API — that lets anyone check content against those disclosures; a compliant third-party tool, clearly linked from the provider’s interface, can be used instead. SB 1000 also removed the older duty to offer a visible (“manifest”) label, and added a temporary exemption for systems designed primarily as assistive technology, which expires on January 1, 2029.

The genuinely hard cases sit at the role boundary. Embedding a third-party system as-is leaves the duties with its producer and makes you a licensee instead — with your own obligations not to strip the disclosure capability and to fix or stop if notified. White-labeling, fine-tuning or modifying that system can turn you into a person who “otherwise produces” it. This checker therefore answers “needs legal review” for those arrangements rather than guessing. In keeping with how this site works, the checker runs entirely in your browser: your answers are never sent to a server — they exist only on this page and, if you use it, in the result link you copy.

How it works

Scope under the California AI Transparency Act turns on one definition. A “covered provider” is “a person that creates, codes, or otherwise produces a generative artificial intelligence system that is publicly accessible within the geographic boundaries of the state” (Business and Professions Code § 22757.1). “Generative AI system” means AI that can generate derived synthetic content — text, images, video and audio — that emulates the structure and characteristics of its training data. As enacted by SB 942 in 2024, the definition also required the system to have “over 1,000,000 monthly visitors or users”; SB 1000 (Chapter 861, Statutes of 2026), an urgency statute effective September 30, 2026, deleted that element, so no size, user or revenue threshold remains. The chapter does not apply to products or services that offer exclusively non-user-generated video game, television, streaming, movie or interactive experiences.

Two duties follow for covered providers, both operative since August 2, 2026 (§ 22757.6, as AB 853 amended it). First, the latent disclosure (§ 22757.3(b)): in image, video or audio content the system creates or alters — beyond “minor modifications” such as brightness or contrast changes, sharpening, resizing, cropping, format conversion or audio denoising — the provider must include, to the extent technically feasible, a disclosure conveying the provider’s name; the system’s name and version number; the time and date of creation or alteration; a unique identifier; and whether the system created or altered the content (a sixth element, whether the system is designed primarily as assistive technology, joins the list in 2029). The disclosure must be detectable by the provider’s verification tool, consistent with widely accepted industry standards, and permanent or extraordinarily difficult to remove. Second, the disclosure verification tool (§ 22757.2, renamed from “AI detection tool” by SB 1000): free and publicly accessible; lets a user assess whether content was created or altered by the provider’s system; outputs detected system provenance data (personal provenance data only with the person’s express consent); accepts uploads and URLs and supports an API; collects user feedback; and strictly limits the collection and retention of personal information from its users. A provider may satisfy the tool duty by clearly directing users to a compliant third-party tool. SB 1000 deleted the original requirement to offer users a manifest (visible) disclosure.

Providers that license their systems must require by contract that licensees maintain the disclosure capability (§ 22757.3(c)). Under the post-2026 mechanics, a provider that knows an identifiable licensee has modified the system into noncompliance must, within 72 hours, terminate the licensee’s authorization or notify the licensee; the licensee must then bring the system into compliance or stop using it and report back within 96 hours, with unresolved cases reported to the Attorney General. Violations carry a civil penalty of $5,000 per violation — each day a discrete violation — in actions by the Attorney General, a city attorney or a county counsel (§ 22757.4); there is no private right of action. SB 1000 temporarily exempts systems designed primarily as assistive technology until January 1, 2029, and makes falsely claiming that exemption a $50,000-per-violation offense. Beyond providers, large online platforms (over 2,000,000 unique monthly users) owe provenance-detection and display duties from January 1, 2027 (§ 22757.3.1), GenAI hosting platforms may not knowingly offer non-compliant systems for download from the same date, and capture-device manufacturers owe default latent-disclosure duties from January 1, 2028.

Worked examples

A 40,000-user AI headshot app: A two-person startup sells AI headshots through a public website; about 40,000 people use it a month. Under the original SB 942 this app was safely under the 1,000,000-user threshold. Since SB 1000 deleted the threshold on September 30, 2026, the startup is a covered provider today: its generated images need latent disclosures carrying all five fields, and it must offer a free verification tool (or link a compliant third-party one). Result pattern: likely in scope.

A retailer embedding a stock image API: An online retailer integrates a well-known third-party image-generation API, unmodified and under the vendor’s name, to create product backgrounds. The retailer does not create, code or produce the system — the vendor does — so the provider duties sit with the vendor. The retailer’s duties are the licensee ones: keep the disclosure capability intact, and if the vendor ever notifies it of noncompliance, fix or stop within the statutory window. Result pattern: likely out of scope as a provider.

A white-label “AI studio”: A marketing platform licenses a foundation image model, fine-tunes it on client brand assets, and sells the output inside a product branded entirely as its own “AI studio”. Whether the platform “otherwise produces” the system it offers is genuinely arguable — its branding and modifications point one way, the licensed core the other. This is exactly the case the checker refuses to guess: needs legal review.

Quick reference

Frequently asked questions

More free tools