AB 3030 vs. SB 942: Which AI Disclosure Law Applies to Your Bot?
Confused by the new alphabet soup of CA laws? We break down the big two. 🥣
The Confusion
California passed multiple AI transparency laws in the same session. The two most prominent are AB 3030 and SB 942. They sound similar, but they cover different things.
AB 3030: The Interaction Law
- Focus: Interactions with people.
- Trigger: When an AI generates text, image, video, or audio that simulates a person.
- Requirement: A disclosure at the start of the interaction (e.g., "I am an AI").
- Target: Chatbots, call center AI, virtual assistants.
SB 942: The Content Law
- Focus: The content itself.
- Trigger: When a "covered provider" (1M+ users) generates synthetic content.
- Requirement: Watermarking (manifest and latent) and a detection tool.
- Target: Image generators (like Midjourney), video generators (like Sora), and large-scale text models.
The Overlap
If your medical chatbot (AB 3030) sends the patient an AI-generated diagram of their heart condition (SB 942), you likely need to comply with both. You must disclose the bot is an AI, and the image must be watermarked.
Conclusion
Map your features to the regulations. Don't assume one covers the other. They are complementary, not mutually exclusive.
Frequently Asked Questions (FAQ)
Which one has higher penalties?
Both are enforced by the Attorney General and can trigger civil penalties. SB 942 has specific daily fines for failing to provide the detection tool.
I have fewer than 1M users. Am I exempt from SB 942?
Yes, SB 942 has a user threshold. AB 3030 does not. Even if you have 1 user, you must comply with AB 3030.
Do these laws apply to B2B tools?
AB 3030 applies if the AI interacts with a "person." A doctor is a person. So yes, it likely applies to provider-facing tools as well.