Last updated: May 10, 2026

AB 3030 vs SB 942: Which California AI Disclosure Law Applies to Your Product?

AB 3030 and SB 942 are both California AI disclosure laws effective January 1, 2026 — but they target completely different compliance obligations. AB 3030 applies to healthcare providers using AI to communicate with patients. SB 942 applies to large AI content platforms distributing AI-generated media to California users. They have different triggers, different requirements, and different penalties. For some products, both apply simultaneously.

AB 3030 applies if:

You are a California healthcare provider
You use generative AI to produce communications
Those communications are sent directly to patients

No user threshold — applies from your first patient

SB 942 applies if:

You generate AI-generated content (text, image, audio, video)
That content is distributed to California users
You have 1M+ monthly California users (for detection tool)

Watermarking applies regardless of platform size

Side-by-Side Comparison

Dimension	AB 3030	SB 942
Full name	Healthcare AI Disclosure Act	California AI Transparency Act
What it covers	Generative AI used to produce clinical communications sent directly to patients	AI-generated content (text, audio, images, video) distributed to California users
Who must comply	California healthcare providers that deploy generative AI for patient-facing communications	Any platform or provider that generates and distributes AI content to California users
User threshold	None — applies from first patient interaction	Watermarking: no threshold. Free AI detection tool: 1 million+ monthly California users
Core requirement	Either (1) a licensed clinician reviews each AI output before sending, OR (2) every AI-generated communication carries a specific disclaimer + human contact instructions	AI-generated content must carry manifest (visible) and latent (embedded) provenance markers. Platforms with 1M+ CA users must also provide a free public AI detection tool
Content types covered	AI-generated text communications to patients (messages, summaries, instructions, follow-ups)	All AI-generated content modalities: text, audio, image, video
Sector	Healthcare only	All sectors — not healthcare-specific
Penalty per violation	Up to $2,500 per violation (per patient communication) + full provider liability for patient harm from unreviewed AI output	AG enforcement under UCL §17200; civil penalties, injunctive relief, restitution. Failure to provide required detection tool: additional daily penalties
Enforced by	California Medical Board; California Attorney General	California Attorney General
Companion laws	Works alongside AB 489 (patient AI identity disclosure)	Companion to AB 3211 (medical image watermarking)
Can both apply?	Yes — if the product is a large healthcare AI platform	Yes — if the product is a large healthcare AI platform

AB 3030 — Healthcare AI Communication Disclosure

AB 3030 is a healthcare-sector law. It governs what happens when a California healthcare provider uses generative AI to produce content that is then sent to a patient. The law does not restrict the use of AI internally — it governs the moment AI output crosses the boundary between the provider and the patient.

What AB 3030 requires

For every AI-generated clinical communication sent to a patient, the healthcare provider must choose one of two compliance paths:

Human review path: A licensed healthcare professional reviews and approves the specific AI output before it is sent. No disclaimer is required if this review occurs.
Disclaimer path: The communication includes a specific disclaimer stating it was AI-generated, was not reviewed by a human provider, and provides instructions for reaching a human provider. No review is required if this disclaimer is included.

What triggers AB 3030

AI-generated post-visit care summaries sent to patients
AI-drafted follow-up messages based on lab results or imaging
Automated AI health coaching messages based on sensor or EHR data
AI-generated prescription refill instructions or medication reminders with clinical context
LLM-written prior authorization explanations sent to patients

What does NOT trigger AB 3030

AI used internally by clinicians (not sent to patients)
Scheduling or administrative AI that does not generate clinical content
Non-generative AI (rule-based systems, decision trees, predictive models that don't generate text)

AB 3030 penalty exposure is per message, not per incident

Each AI-generated patient communication lacking the required review or disclaimer is a separate $2,500 violation. A health system sending 5,000 AI-generated care gap messages per day without compliant disclosures faces $12.5 million in potential daily penalty exposure — not a one-time fine.

SB 942 — AI Content Provenance and Watermarking

SB 942 is a broad AI content transparency law that applies across all sectors, not just healthcare. It governs the provenance of AI-generated content — ensuring that users can identify when content was created by AI, regardless of the context in which it is received.

What SB 942 requires

SB 942 imposes two distinct requirements with different thresholds:

Watermarking (all sizes): Any AI-generated content — text, audio, images, video — distributed to California users must carry both manifest (visible, human-readable) and latent (embedded, machine-readable) provenance markers identifying the content as AI-generated.
Detection tool (1M+ California users): Platforms with 1 million or more monthly California users must provide a free, publicly accessible AI content detection tool that allows anyone to verify whether a given piece of content was AI-generated.

What triggers SB 942

AI-generated images produced by healthcare platforms and shared with patients (diagnostic visualizations, anatomical illustrations, AI-rendered imaging previews)
AI-generated audio (including voice synthesis used in telehealth platforms)
AI-generated video content distributed through health information platforms
AI-generated text on platforms that meet the size threshold

The 1 million user threshold — what it means

The threshold applies to California users, not total global users. A national digital health platform with 8 million total users but only 700,000 California monthly active users does not trigger the detection tool requirement. However, all platforms — regardless of size — must apply provenance watermarking to AI-generated content distributed to California users.

When Both Laws Apply Simultaneously

A telehealth platform or healthcare AI system that both communicates with patients via AI-generated messages (AB 3030) and distributes AI-generated media content to a large California user base (SB 942) must comply with both laws independently. Compliance with one does not satisfy the other.

Practical examples of dual-trigger scenarios:

A large telehealth platform that sends AI-generated care summaries to patients (AB 3030) and generates AI anatomy images or AI-narrated educational videos for patients (SB 942, and potentially the detection tool requirement)
A hospital patient portal that generates AI follow-up messages (AB 3030) and uses AI to create patient-facing health infographics or discharge illustration images (SB 942)
A direct-to-consumer digital health app with 2M+ California users that generates both AI health coaching text (AB 3030 if clinical) and AI-synthesized audio explanations (SB 942)

Which Law Applies to Your Product

Product / Use Case	AB 3030	SB 942 (watermark)	SB 942 (detection tool)
Small clinic using AI to draft patient follow-up messages	✓ Yes	Only if AI images distributed	No (below threshold)
AI image generator used by radiologists (not patient-facing)	No (not patient-facing)	✓ Yes	Only if 1M+ CA users
Large telehealth platform (3M CA users) with AI patient messaging + AI health video	✓ Yes	✓ Yes	✓ Yes
AI medical content platform with 2M CA users (no direct patient care)	Probably not (no clinical communication to patients)	✓ Yes	✓ Yes
Hospital AI that assists clinicians only, never generates patient-facing content	No	No	No

How to Comply with Each Law

AB 3030 compliance steps

Inventory every AI-generated communication sent to patients — by type (care summaries, follow-ups, medication reminders, etc.)
For each type, choose a compliance path: human review before sending OR disclaimer on automated output
If using the disclaimer path, generate compliant disclaimer language that identifies AI generation, absence of human review, and human contact instructions
Build audit logs capturing AI output, reviewer identity (if review path), approval decision, and timestamp
Document the chosen path for each communication type in a written AI usage policy

SB 942 compliance steps

Audit all AI-generated content distributed to California users (text, audio, images, video)
Implement manifest watermarking: visible label on AI-generated content identifying it as AI-generated
Implement latent watermarking: embedded C2PA-compatible metadata in AI-generated content files
If you have 1M+ monthly California users: build or integrate a free, publicly accessible AI content detection tool
Establish a process to maintain watermarking as the product evolves and new AI content types are introduced

AB 3030 Disclosure Generator →

Free tool: generate compliant disclaimer language for AB 3030 patient communications. Ready to deploy in 60 seconds.

SB 942 Compliance Guide →

Complete resource hub: watermarking requirements, manifest vs latent disclosures, and the 1M user detection tool obligation.

Frequently Asked Questions

What is the difference between AB 3030 and SB 942?

AB 3030 applies to California healthcare providers that use generative AI to produce communications sent directly to patients — it requires either human clinical review of each AI output or a specific disclaimer on every AI-generated message. SB 942 applies to AI content platforms with 1 million or more monthly California users — it requires AI-generated content to carry manifest (visible) and latent (embedded) provenance markers, and it requires those large platforms to provide a free public AI detection tool. AB 3030 is a healthcare-specific law; SB 942 is a broad AI content transparency law.

Can AB 3030 and SB 942 both apply to the same product?

Yes. A large telehealth platform that uses generative AI to produce patient communications and has 1 million or more monthly California users must comply with both laws simultaneously. AB 3030 governs the disclosure and oversight of the patient-facing AI communications; SB 942 governs the provenance watermarking of any AI-generated content (images, audio, video, text) distributed through the platform.

Does the SB 942 1 million user threshold apply to California users only?

Yes. The SB 942 threshold is 1 million monthly California users — not total global users. A platform with 10 million total users but only 800,000 California users does not trigger SB 942's detection tool requirement. However, the watermarking obligations under SB 942 apply to AI content distributed to California users regardless of the platform's total size.

Which law has higher penalties — AB 3030 or SB 942?

AB 3030 carries penalties of up to $2,500 per violation — meaning per patient communication that lacks the required disclosure or human review. For high-volume platforms, this creates very large aggregate exposure because each individual AI-generated message is a separate violation event. SB 942 penalties are enforced by the California AG under the Unfair Competition Law; violations can result in civil penalties, injunctive relief, and restitution. AB 3030 also uniquely makes the healthcare provider fully liable for any patient harm caused by unreviewed AI output, with no statutory safe harbor.

Does AB 3030 apply to AI used only by clinicians, not patients?

No. AB 3030 specifically applies to generative AI that produces communications sent directly to patients. An AI clinical decision support tool that presents recommendations to physicians — and whose outputs are never sent to patients — is not within AB 3030's scope. The law is triggered at the point where AI-generated content reaches the patient.

Does SB 942 apply to text-based AI, or only images and video?

SB 942 applies to AI-generated audio, images, video, and text — all modalities are covered. For platforms that generate AI text content distributed to California users, both manifest (visible label) and latent (embedded metadata) watermarking is required. The law does not carve out text-only platforms.

If I comply with AB 3030, am I also compliant with SB 942?

No. AB 3030 compliance (human review or patient disclaimer on AI clinical communications) does not address SB 942's provenance watermarking or detection tool requirements. These are separate obligations with separate technical and procedural implementations. If both laws apply to your product, you must satisfy each independently.