Presentation on AI in health care prompts board to form ad hoc AI committee to study AB 489

A public presentation and sustained public comment on artificial intelligence in health care led the California Board of Occupational Therapy on Feb. 27 to authorize formation of an ad hoc committee to examine how newly enacted and pending state laws should be applied to occupational‑therapy practice.

Tammy Richmond, who identified herself as an occupational therapist, chair of an American Telemedicine Association telerehabilitation special interest group and founder of a hybrid telehealth company, told the board that AI is fundamentally different from telehealth and is increasingly being embedded in clinical tools. “AI is not telehealth,” Richmond said during her presentation. She described two common categories — generative AI (chatbots and natural‑language systems) and predictive AI (pattern‑based forecasting) — and warned that "agentic" AI can observe, decide and act without a human in the loop.

Richmond offered concrete examples in which hospitals or vendors could deploy apps or digital therapeutics that effectively perform components of occupational therapy and bill under recent remote therapeutic monitoring (RTM) codes. She said those codes were revised to require only limited data collection (two days), enabling non‑clinical actors to use them to bill for services that previously required more clinician involvement. Richmond warned this could leave occupational therapists and assistants out of clinical loops and erode professional roles: "They'll use the RTM codes, and that's what's scary for me because it's true. They will use them."

Board members and attendees pressed her for examples and for policy options. Several public commenters — including academic fieldwork coordinators and representatives from the Occupational Therapy Association of California and SEIU UnitedHealthcare Workers West — thanked Richmond and offered to serve on an ad hoc committee to study regulatory language, clinical safeguards and consumer protections.

The board discussed the difference between regulation and enforcement when the AI model operator is not a licensee; staff noted that the board’s authority is limited when an unlicensed vendor deploys an AI model but the department (DCA) may have investigatory authority under new statutory language. Staff recommended the ad hoc committee scope regulatory language, consider transparency requirements and consult legal counsel about enforcement pathways under AB 489 and related bills.

Board members volunteered to lead the committee and public and professional volunteers offered to participate. The board’s president said staff will follow up to finalize membership and confirm procedural details and quorum rules. No formal regulatory language was adopted at the meeting; the committee will return with recommendations.

The board’s decision to form an ad hoc AI committee follows concerns raised in the presentation about patient safety, biased data sets, clinical oversight and liability. Richmond urged the board to pursue "model language" and to prioritize areas of practice where agentic AI should never act without a licensed clinician’s oversight.