A recent U.S. federal proposal would deregulate transparency requirements for artificial intelligence tools used in health care, significantly changing how these technologies are vetted before use. Previously, developers had to disclose detailed information about how clinical AI products were developed and tested. The new plan would remove this requirement, meaning health care providers themselves would have to demand and evaluate that information directly from AI vendors to ensure safety and effectiveness for their patient populations. Experts warn this could make it harder for hospitals and clinics to reliably compare products and choose the right tools for clinical use.
Under the proposed rule from the Department of Health and Human Services’ health IT office, a host of “model card” transparency requirements—which act like nutrition labels for AI, revealing how models are trained and evaluated—would be eliminated from certification standards. This rollback is part of a broader deregulatory effort aimed at reducing compliance burdens on developers and promoting innovation. Supporters argue that trimming these rules will accelerate AI adoption by removing hurdles that slow product rollout and increase development costs.
Critics, however, say that shifting the verification burden onto health systems raises risks. Hospitals and clinics already operate under tight resource constraints, and many lack the technical expertise or staffing to thoroughly interrogate complex AI tools. Without mandated disclosures from developers, providers could struggle to identify biases, safety issues, or limitations in AI products, potentially affecting patient care quality and outcomes. The change also narrows the transparency that clinicians and patients rely on to trust AI recommendations.
This proposal reflects larger regulatory trends in healthcare AI, where policymakers are debating how to balance innovation and safety. Some industry stakeholders welcome reduced federal oversight, saying it encourages technological advancement, while others call for frameworks that still ensure accountability and protect patients. As the rule advances, these competing priorities will shape how AI tools are evaluated and implemented across U.S. health systems.
