/ 3 min read / Reed Smith Client Alerts

California AG challenges federal move to roll back AI transparency in health care

On February 27, California Attorney General Rob Bonta issued a letter commenting on a proposed rule released by the U.S. Department of Health and Human Services (HHS) and the Assistant Secretary for Technology Policy/Office of the National Coordinator for Health Information Technology (ASTP/ONC).

HHS and ASTP/ONC published the proposed rule, titled “Health Data, Technology, and Interoperability: ASTP/ONC Deregulatory Actions To Unleash Prosperity,” in the Federal Register on December 29, 2025, seeking to revise or eliminate many of the current health IT certification criteria designed to protect patient privacy and ensure data accuracy. In the proposed rule, the federal government seeks to remove these criteria to encourage the development and adoption of AI in health IT systems. 

Attorney General Bonta’s response focuses primarily on the rule’s proposed removal of the model card requirement previously implemented by the Biden Administration. A model card, which Bonta likens to a “nutrition label” for health IT systems, describes how a certain AI model was developed and provides instructions for how it should be used. Requiring model cards is a transparency measure that proponents argue helps users and supervisors combat algorithmic bias when using AI systems. 

Attorney General Bonta contends that AI transparency is crucial to combating algorithmic discrimination, and that model cards are crucial to transparency. He points to California studies examining the adverse impact of algorithmic bias on patients and argues that those studies reveal the need for guardrails in health care AI. Attorney General Bonta also challenges the federal government’s basis for removing the requirement. ASTP/ONC cites various Executive Orders to justify its actions, as well as four factors supporting its proposal to eliminate the model card certification criteria: (1) lack of provider time to engage with model cards, (2) lack of perceived benefit of transparency, (3) lack of a model card requirement for other tech companies, and (4) promotion of progress by removing the burdensome model card requirement. Attorney General Bonta challenges these bases, arguing that the removal of the model card requirement will make it harder for health care providers to comply with other state and federal laws and will hinder innovation in health IT, contending that transparency encourages development.

This is not the first time Attorney General Bonta’s office has weighed in on AI regulation in health care. Last year, Bonta issued an advisory addressing the application of existing California health care laws to AI, providing guidance to health care entities about their current obligations under state law. This latest effort underscores a growing tension between state regulators such as AG Bonta, who favor safety, accountability, and transparency, and the current federal administration, which favors flexibility, self-regulation, and accelerated AI adoption. This tension follows the Trump Administration’s release of an executive order in December 2025 seeking to preempt state AI regulation, setting the stage for an ongoing conflict over who will define the guardrails for AI in health care.

As the AI health care regulatory landscape continues to mature, this federal–state tension will shape how health care entities develop and deploy AI tools. Reed Smith attorneys are closely monitoring these developments and are available to help organizations navigate the evolving regulatory environment.

Client Alert 2026-57

Related Insights