Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Health Ai topic
No spam. Unsubscribe anytime.
Senate health committee advances bill to require AI guardrails in patient care
Summary
AB 2575 would require employers to give health workers information about AI tools, protect clinicians who override AI recommendations, and prevent developers or employers from shifting liability entirely to frontline workers; the health committee voted to pass the bill 8‑2 after supporters and multiple health‑industry groups debated costs and accountability.
Get email alerts on the Health Ai topic
No spam. Unsubscribe anytime.
The Senate Committee on Health voted to pass AB 2575, a measure from Assemblymember Ortega that would set statewide guardrails for high‑risk artificial intelligence used in patient care.
The bill requires employers to provide clinicians access to key information about an AI tool’s limitations and intended uses, prohibits retaliation against health workers who, in good faith, override or follow AI recommendations based on professional judgment, and prevents developers or employers from avoiding liability by shifting all blame to clinicians. "In healthcare, artificial intelligence should support clinical judgment, but never replace it," Assemblymember Ortega said when she introduced the bill, and she told senators she accepted committee amendments that narrow disclosure requirements and remove a 90‑day notice provision.
Nurses and labor groups strongly supported the measure. "When technology is used in patient care, we want to understand it, question it, and intervene to protect our patients," Sandy Redding, president of the California Nurses Association, testified, urging senators for an aye vote. Sarah Flocks of the California Labor Federation said standard, statewide guardrails were necessary even when employers assert a "human in the loop" policy.
Medical providers and industry groups warned the committee that the measure, as written, could raise costs and operational uncertainty and discourage deployment of AI tools now used for radiology review, risk detection and care coordination. "This bill threatens these benefits by creating requirements that will cause clinicians to scale back or delay the use of proven technology," George Soares of the California Medical Association argued in opposition, citing potential strain on hospitals and patient access.
Committee members discussed bias, clinical responsibility, and the scope of the amendments. Senator Durazo acknowledged the need to avoid removing accountability when clinical judgment is wrong and described personal examples of unconscious bias in care; Ortega said she will continue discussions with stakeholders.
The committee motion to pass AB 2575 as amended was made and the measure was placed on call; the roll call recorded an 8‑2 favorable tally. The bill will next be referred to the Committee on Labor, Public Employment, and Retirement.
