Health plans would need human judgment for medical coverage denials

Group health plans and insurers offering group or individual coverage would have qualified, licensed health professionals independently decide denials involving medical judgment. This would cover initial decisions and internal appeals based partly or fully on medical judgment, such as whether treatment is needed or effective. AI could assist, but it could not issue or control the decision or replace the professional's independent judgment. The rules would apply to plan years starting on or after January 1 of the first calendar year beginning at least 12 months after enactment.

Group health plans and insurers offering group or individual coverage would disclose AI use in denial notices and provide review records on request. For initial denials and internal appeals involving medical judgment, plans would keep records identifying the AI system, its role, and its outputs. When AI was used, the notice would explain its role and identify the deciding professional's name, license, and credentials. The rules would apply to plan years starting on or after January 1 of the first calendar year beginning at least 12 months after enactment.

AI use in reviews of care would count as a treatment limit under rules comparing mental health and substance use coverage with medical and surgical coverage. The comparisons would cover both written coverage rules and how they work in practice. When AI shaped a treatment limit, information about its function and effects would have to let federal officials compare coverage. The rules would apply to plan years starting on or after January 1 of the first calendar year beginning at least 12 months after enactment.