California’s AI Transparency Act is real law, but it is not a blanket requirement that every person visibly label every AI-generated image, video, or text item. SB 942 was approved on September 19, 2024, became effective January 1, 2025, and its core provisions became operative on August 2, 2026. The law primarily regulates qualifying providers of publicly accessible generative-AI systems, with additional requirements for certain platforms and device manufacturers taking effect later.
Its central tools are provenance disclosures, a free provider-operated verification tool, licensing controls, and privacy protections. The law focuses principally on images, video, audio, and combinations of those formats—not standalone text in the same way.
![Claude AI for Beginners Bible: [5 in 1] The Ultimate Guide to Automate Your Work, Save Hours Every Week, and Use AI for Real-World Results](https://m.media-amazon.com/images/I/415+fSJacsL._SL160_.jpg)
As an Amazon Associate I earn from qualifying purchases.
What is SB 942?
SB 942 is formally called the California AI Transparency Act. It was enacted as Chapter 291, Statutes of 2024, and added provisions beginning at Section 22757 of the California Business and Professions Code. Its stated purpose is to help people inspect digital provenance and better determine whether media was generated or altered by artificial intelligence.
The law is often described as an AI “watermarking” law. That shorthand is incomplete. SB 942 combines:
#1 Best Overall
- A free public detection or verification tool operated by covered AI providers.
- An option for users to attach a visible, human-perceivable disclosure.
- Required hidden or machine-detectable provenance information.
- Contractual safeguards for licensed AI systems.
- Privacy restrictions on verification tools.
- Later duties for certain online platforms, hosting platforms, and capture-device manufacturers.
The official statute is available through the California Business and Professions Code.
When does SB 942 apply?
| Date | Event |
|---|---|
| September 19, 2024 | Governor Gavin Newsom approved SB 942. |
| January 1, 2025 | The enacted chapter became effective. |
| January 1, 2026 | Later AB 853 amendments became effective, although the core chapter’s operative date remained delayed. |
| August 2, 2026 | The core provider disclosure and verification provisions became operative. |
| January 1, 2027 | Principal requirements for large online platforms and GenAI hosting platforms become operative. |
| January 1, 2028 | Certain capture-device manufacturer requirements become operative. |
“Effective” and “operative” are not interchangeable. The January 1, 2025 date describes when the law entered the statute book. August 2, 2026 is the key date for the core compliance provisions. Older summaries that say SB 942 began applying on January 1, 2026 may be outdated.
Who is a covered provider?
Under the enacted SB 942 framework, a covered provider is generally a person that:
- Creates, codes, or otherwise produces a generative-AI system;
- Has more than 1 million monthly visitors or users; and
- Makes the system publicly accessible within California.
The law is not limited to companies headquartered in California. The important geographic question is whether the system is publicly accessible within the state.
The statute does not provide a simple public formula for counting monthly visitors or users. Questions may arise when a system is offered through multiple websites, APIs, resellers, embedded services, or interfaces. The meaning of “publicly accessible” may also require analysis for global services that do not specifically target California.
The threshold applies to providers of qualifying systems, not automatically to every person who uses an AI product. A consumer, freelancer, employee, or small business that merely uses a covered system generally is not transformed into a covered provider by doing so.
What content is covered?
The principal disclosure provisions cover:
- Images;
- Video;
- Audio; and
- Content combining those media types.
Generative AI is defined broadly enough to include synthetic text, but Section 22757.3’s manifest and latent disclosure obligations specifically identify image, video, audio, and combinations of those formats. Standalone text should not be presented as subject to the same disclosure regime without qualification.
Free tools Windows power users keep installed
One-click scans. No signup required.
What must covered providers do?
| Requirement | Who it affects | What it means |
|---|---|---|
| Free detection tool | Covered provider | Users must be able to assess whether qualifying media was created or altered by that provider’s system. |
| Manifest disclosure | Covered provider | The provider must offer users an option for a clear, visible disclosure identifying qualifying content as AI-generated. |
| Latent disclosure | Covered provider | Qualifying generated or altered media must contain hidden or machine-detectable provenance information when technically feasible and reasonable. |
| License controls | Provider and licensee | A license must preserve the system’s disclosure capability. If the provider discovers that a licensee disabled it, the provider must revoke the license within 96 hours. |
| Privacy safeguards | Detection-tool provider | The tool must limit collection and retention of personal information, submitted content, and personal provenance data. |
Manifest disclosures
A manifest disclosure is the visible or otherwise human-perceivable notice attached to generated or altered media. The provider must offer the user the option to include it. The disclosure must identify the content as AI-generated, be clear and conspicuous, suit the medium, and be understandable to a reasonable person.
It must be permanent or extraordinarily difficult to remove to the extent technically feasible. The statute does not prescribe one exact word, icon, logo, or design.
Most importantly, the provider’s duty to offer the option is not phrased as a universal duty requiring every user to activate a visible label on every output. A provider may make the option available, while the user’s own obligations may arise from other laws, contracts, platform rules, advertising requirements, or professional standards.
Latent disclosures
A latent disclosure is hidden or machine-detectable provenance information embedded in or attached to media. To the extent technically feasible and reasonable, it must convey:
- The provider’s name;
- The name and version number of the generative-AI system;
- The time and date of creation or alteration; and
- A unique identifier.
The disclosure must be detectable by the provider’s tool, follow widely accepted industry standards, and be permanent or extraordinarily difficult to remove to the extent technically feasible.
Provenance is not proof of truth. It may indicate where content came from or how it was modified, but it does not establish that a caption is accurate, that an image depicts a real event, or that a person consented to the content.
The free detection tool
A covered provider’s tool must be available at no cost and must:
- Assess whether image, video, audio, or multimedia content was created or altered by that provider’s system;
- Output detected system provenance data without outputting personal provenance data;
- Be publicly accessible, subject to reasonable security restrictions;
- Accept uploaded content or a URL linking to online content; and
- Provide an API so users can invoke the tool without visiting the provider’s website.
The provider must also collect feedback about the tool’s effectiveness and use relevant feedback in efforts to improve it.
Recommended Free Tools
This is not necessarily a universal AI detector. A provider’s tool is principally required to identify content associated with that provider’s own system. A negative result does not prove that a file was created by a human or by no AI system.
Privacy rules
The provider generally may not collect or retain personal information from detection-tool users, retain submitted content longer than necessary to comply with the statute, or retain personal provenance data from submitted content.
There is a limited exception for contact information supplied by users who submit feedback and opt in to follow-up contact. That information may be used only to evaluate and improve the tool’s effectiveness.
Users should still review the particular provider’s privacy documentation before uploading sensitive, unpublished, personal, or commercially confidential media. The statute’s restrictions do not automatically mean that every third-party AI detector has the same privacy protections.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
What happens when a provider licenses its model?
If a provider licenses its generative-AI system to another party, the provider must contractually require the licensee to maintain the system’s ability to include the required latent disclosure.
If the provider discovers that the licensee modified the system so it can no longer include the disclosure, the provider must revoke the license within 96 hours of discovering the modification. The licensee must stop using the licensed system after revocation.
This creates separate compliance questions for the model provider, the downstream licensee, and an ordinary user of the resulting product.
Later rules for platforms and devices
AB 853, Chapter 674, Statutes of 2025, added requirements for additional parts of the digital-content ecosystem. Those obligations do not all become operative on the same date as the core provider provisions.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteLarge online platforms: January 1, 2027
A large online platform must generally be able to detect compliant provenance data embedded in or attached to distributed content. It must provide an interface showing whether provenance data is available, display information about authenticity, origin, or modification history, and indicate the relevant AI system or capture device when applicable.
The platform must also indicate whether digital signatures are available, allow users to inspect available compliant provenance data, and avoid knowingly stripping compliant provenance data or digital signatures when technically feasible.
GenAI hosting platforms: January 1, 2027
A GenAI system hosting platform may not knowingly make available a generative-AI system that does not place the required disclosures under Section 22757.3.
Capture-device manufacturers: January 1, 2028
For covered capture devices first produced for sale in California on or after January 1, 2028, manufacturers must provide an option to include latent disclosure, include latent disclosures by default, and provide specified manufacturer, device, and time-and-date information, subject to technical feasibility and recognized standards.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →What SB 942 does not require
- It does not generally require every California resident to label every AI-assisted work. The main duties fall on qualifying providers and, later, specified platforms and manufacturers.
- It does not impose the same disclosure regime on standalone text. The core manifest and latent disclosure provisions focus on image, video, audio, and combinations of those formats.
- It does not guarantee that disclosures survive every workflow. Screenshots, cropping, re-encoding, editing, and unsupported software can remove or obscure provenance.
- It does not prove authenticity or accuracy. Provenance identifies origin or modification information, not whether the content’s claims are true.
- It does not automatically fine ordinary users. The statute creates a civil-enforcement mechanism aimed at violations by covered entities.
- It does not make every platform immediately responsible. Many platform and device provisions have later operative dates.
Enforcement and penalties
A violation may carry a $5,000 civil penalty per violation. The California Attorney General, a city attorney, or a county counsel may bring a civil action. A prevailing plaintiff may recover reasonable attorney’s fees and costs.
Under the current code text, each day that a covered provider, large online platform, or capture-device manufacturer remains in violation is treated as a separate violation. That means potential exposure is not necessarily limited to one $5,000 amount.
For a licensee that continues using a system after its license has been revoked, the statute authorizes actions for injunctive relief and reasonable attorney’s fees and costs.
The statute does not establish an automatic private damages claim for every consumer who encounters unlabeled or incorrectly labeled content. Actual enforcement interpretation may depend on future agency and court action.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Practical examples
A large image generator creates an image
If the provider meets the covered-provider definition and makes its system publicly accessible in California, its image workflow may need to support the required latent disclosure, offer an optional manifest disclosure, and provide the free detection tool.
A small locally run model generates an image
The original coverage test depends on the provider’s status, public accessibility, and more-than-1-million monthly visitor or user threshold. A small system may not meet that threshold under the enacted framework, although other laws or future amendments could still matter.
A social platform receives a provenance-bearing image
The platform’s obligations depend on whether it qualifies as a large online platform and on the applicable operative date. The later platform rules are principally scheduled for January 1, 2027, not August 2, 2026.
A user uploads a suspicious video to a provider’s tool
The tool may report whether the provider’s own system created or altered the video and may return system provenance data. A negative result does not establish that the video is human-made.
A licensee disables provenance support
Once the provider discovers the modification, the provider must revoke the license within 96 hours. The licensee must stop using the licensed system after revocation.
A screenshot is reposted
A screenshot may remove embedded provenance and may also omit a visible label. A missing disclosure is therefore not conclusive evidence that content was created without AI or that a provider violated the law.
A text-only chatbot produces an answer
SB 942’s central manifest and latent disclosure provisions should not automatically be described as requiring a label on every standalone text response. The precise obligations depend on the statutory provision and the system’s broader features.
Important failure modes and limitations
- Metadata loss: Screenshots, re-encoding, cropping, and reposting can remove provenance.
- Multiple-model workflows: One system may generate media while another edits it, leaving an incomplete history.
- Human editing: The legal and technical treatment of substantially altered or mixed-origin content may require case-specific analysis.
- Open-source and local models: Provider identity, public accessibility, and user-count calculations may be difficult to determine.
- False negatives: A tool that finds no provenance does not prove human creation.
- Privacy exposure: Users may upload confidential media despite retention restrictions and should review the provider’s terms.
Pending change: SB 1000
As of August 16, 2026, SB 1000 was not enacted law. The bill had passed the Senate and was moving through the Assembly; the official status page listed it for third reading on the Assembly file dated August 17.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →The amended proposal would substantially revise the framework, including potentially:
- Removing the more-than-1-million-user threshold;
- Replacing “AI detection tool” with “disclosure verification tool”;
- Deleting the provider’s obligation to offer users an optional manifest disclosure; and
- Revising the latent-disclosure language.
Those proposals should not be blended into the current SB 942 requirements. The compliance analysis may change if SB 1000 is later enacted. The official SB 1000 text and status page provide the relevant legislative information.
A practical coverage checklist
- Are you merely using an AI tool? SB 942 generally does not impose a direct labeling duty on ordinary users solely because they use AI.
- Do you provide or operate a generative-AI system? Review the provider definition, public accessibility, California availability, and user threshold.
- Does the system generate or alter image, video, audio, or combined media? Those formats receive the main manifest, latent-disclosure, and detection-tool treatment.
- Do you license the system? Review contracts, provenance capability, monitoring, and the 96-hour revocation requirement.
- Do you operate a large online platform or GenAI hosting platform? Assess the later provisions scheduled principally for 2027.
- Do you manufacture qualifying capture devices? Review the requirements scheduled for 2028.
Organizations should also document model versions, timestamps, unique identifiers, provenance standards, API controls, retention periods, user notices, license audits, and procedures for responding to a disabled disclosure capability. Because the statute uses phrases such as “technically feasible” and “extraordinarily difficult to remove,” legal and engineering teams may need to assess implementation choices together.
The Bottom Line
Bottom line: SB 942 is best understood as a provider-focused provenance and verification regime, not a blanket consumer watermarking law. Its core provisions became operative August 2, 2026 and primarily concern qualifying providers of publicly accessible generative-AI systems that create or alter images, video, audio, or multimedia. Users should not assume that a label proves truth, that an unlabeled file is human-made, or that pending SB 1000 proposals are already law.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
