AI/ML Medical Device Documentation Requirements in California (AB 3030)
California AB 3030 requires AI/ML medical device documentation to include a clear, prominent disclosure that the device is AI — at the start of every patient interaction, in plain language, with a visible path to a human. This 2026 checklist gives you the four documentation elements California regulators audit for.
Who Needs to Comply?
AB 3030 applies broadly to any "generative artificial intelligence" system that interacts with a person. If your device is specifically a chatbot, see our dedicated guide: Is your medical chatbot legal under California AB 3030? — it covers the same disclosure rules with chatbot-specific UX guidance and the human-review exemption. In the context of medical devices, this includes:
- Chatbots: Any conversational agent used for triage, scheduling, or mental health support.
- Generated Reports: Summaries of patient data, lab results, or imaging reports created by GenAI.
- Virtual Assistants: Avatars or voice assistants that guide patients through therapy or medication adherence.
If your device generates text, images, video, or audio that simulates human communication, you are likely covered.
The Checklist
Compliance isn't complicated, but it must be precise. Use this checklist to audit your device:
- Prominent Disclosure: Is there a clear statement at the start of the interaction? It cannot be hidden in a "More Info" menu.
- Clear Language: Do you use plain English? Avoid technical jargon. Say "I am an AI assistant," not "I am a large language model based on transformer architecture."
- Instructions: Do you tell the patient how to reach a human? The law requires a clear path to human escalation if the AI cannot resolve the issue.
- Visual Cues: If using an avatar, ensure it does not look deceptively human (see AB 489).
🚀 Pro-Tip: Automate Your Compliance
Don't guess what your disclosure should say. Use our free generator to create legally-compliant text for your app in seconds.
Generate AB 3030 Disclosure NowDesign Tips
Don't bury the disclosure in the Terms of Service. It must be visible in the UI. For chat interfaces, it should be the very first message bubble. For voice interfaces, it should be the first sentence spoken.
Conclusion
Transparency builds trust. Patients are more likely to engage with an AI tool if they know exactly what it is and what its limitations are. Make sure your device is honest about its nature.