A new California law quietly imposes a significant compliance obligation on small medical practices using AI—and most owners haven't heard about it yet.
Most California health care owners don't realize that a provision buried in AB2575 will require them to document and disclose every artificial intelligence tool they use in clinical decision-making. For small physician offices, clinics, and group practices, this creates a new, ongoing compliance responsibility with real documentation demands.
Effective July 1, 2027, any small clinic or physician's office using AI-assisted clinical tools must compile and maintain a formal inventory of those systems. This isn't a one-time checkbox. You'll need to update it annually and keep it current as your practice's AI use evolves.
But the inventory requirement is only the beginning. Under Section 22759(a) of the California Business and Professions Code, you must also be prepared to disclose detailed technical information about each AI system on request. This includes:
These disclosures must be provided separately for each tool and represent a substantive documentation burden—especially for practices that may not have ready access to this technical detail from vendors.
The rule applies to small physician's offices, clinics, and group practices. If your practice uses any AI system to assist in clinical decision-making—whether it's diagnostic imaging analysis, clinical note generation, risk stratification, or treatment recommendations—you're in scope.
This includes tools you may not think of as "AI systems" in the formal sense. If a vendor describes a clinical tool as using machine learning, algorithmic analysis, or predictive modeling, it likely falls under this requirement.
By July 1, 2027: You must have your inventory of AI clinical tools compiled and in place.
Starting July 1, 2027: You must be prepared to disclose the technical details listed above on request, and you must notify relevant parties annually about your AI systems and their characteristics.
This gives practices roughly two and a half years to identify all AI systems in use, gather technical documentation from vendors, and establish processes for maintaining and updating the inventory annually.
Start by auditing your current technology stack. Document every clinical tool that uses AI or algorithmic decision support. Then contact your vendors to request the technical information you'll need: training data sources, known performance gaps or biases, and validation data.
Many vendors may not have this information readily available in the format the law requires. Getting it now—before the deadline—gives you time to work with them and avoid last-minute scrambling.
You'll also want to establish a process for keeping your inventory current as you adopt new tools or retire old ones, since the annual update requirement is ongoing.
For a detailed, practice-specific guide to AB2575's AI inventory and disclosure requirements, contact your local health care trade association or legal counsel.