Right to Override Act

Bill Number

S. 2997

Origin Chamber

Senate

Congress

119th Congress, Session 1

Policy Area

Health

Status

Introduced

Latest Action

2025-10-09: Read twice and referred to the Committee on Health, Education, Labor, and Pensions.

Last Updated

2026-07-08T16:02:12Z

AI-Generated Summary

Purpose of the Legislation

The "Right to Override Act" aims to safeguard the independent clinical judgment of health care professionals when they choose to override recommendations from artificial intelligence clinical decision support systems (AI/CDSS). These systems use algorithms to suggest decisions in patient care, such as diagnoses or treatments. The law ensures that AI tools support, but do not replace, human expertise, while protecting professionals from workplace retaliation for prioritizing patient needs or legal requirements.

Key Provisions

• Definitions (Section 3): Establishes clear terms, including:
• AI/CDSS: Technology using algorithms based on clinical guidelines or data to produce predictions, recommendations, or analyses for health care decisions.
• Covered entity: Employers or entities engaging health care workers, such as hospitals, clinics, schools with nurse offices, or health insurance plans.
• Health care professional: Licensed or certified individuals providing services like diagnosis, treatment, or insurance coverage decisions, including home care providers.
• Override: Making a decision contrary to an AI/CDSS output.
• Adverse employment action: Broadly defined to include firings, demotions, schedule changes, denied promotions, or any action worsening a worker's job conditions.

• Title I: Policies for Using and Overriding AI/CDSS (Sections 101–103):
• Covered entities using AI/CDSS must adopt policies ensuring AI does not override professionals' judgment; allow timely overrides if deemed appropriate for patients or to follow laws (e.g., civil rights); permit feedback on AI issues like biases; and prohibit sharing override data that identifies specific workers or inferable groups.
• Entities must inform and train workers on AI use, override processes, limitations, and biases.
• Requires forming an AI/CDSS committee with balanced representation (including non-managers and unions) to advise on policies, review implementations, and suggest improvements; must meet quarterly.
• Enforcement by the Secretary of Health and Human Services (HHS) through investigations and civil penalties, similar to HIPAA processes; no criminal referrals.
• HHS may issue regulations, consulting with the Department of Labor (DOL) and other agencies.

• Title II: Adverse Employment Actions and Whistleblower Protections (Sections 201–204):
• Prohibits covered entities from taking adverse actions against professionals for lawful overrides.
• Bans retaliation (e.g., threats, harassment) against workers for exercising rights, filing complaints, seeking help, or participating in investigations related to the Act.
• DOL enforces via investigations, subpoenas, reports, and penalties up to $76,987 per violation (or $769,870 for repeats), adjusted annually for inflation; allows union representatives to assist in complaints.
• Provides a private right of action for affected individuals, including damages (actual, treble, or statutory up to $100,000), injunctive relief, attorney fees, and temporary reinstatement; invalidates predispute arbitration or class action waivers.
• Waives state sovereign immunity for suits involving federal funding, allowing actions against state officials for injunctive relief.
• DOL may issue regulations, consulting with HHS and other agencies.

• Title III: General Provisions (Sections 301–304):
• HHS and DOL must develop educational materials within one year on compliance and worker rights.
• Allows state attorneys general or privacy regulators to sue on behalf of residents for violations, seeking injunctions, damages, or fees; federal agencies can intervene.
• Clarifies the Act does not shield professionals from medical malpractice claims for overrides.
• Does not preempt state laws or union agreements.

Significant Changes to Existing Law

This legislation introduces new federal requirements specifically for AI in health care, which were not previously addressed in U.S. law. It builds on existing frameworks like HIPAA (for HHS enforcement), the Fair Labor Standards Act (for DOL investigations), and whistleblower protections (e.g., under aviation law for burdens of proof), but adds unique mandates for AI override policies, worker committees, data privacy on overrides, and anti-retaliation rules tailored to clinical decisions. It also explicitly invalidates arbitration agreements for these disputes, diverging from some employment laws that permit them.

Potential Impacts

• On Government Agencies: HHS and DOL gain expanded enforcement roles, including investigations, penalties, and rulemaking, potentially increasing administrative workloads. States can participate in enforcement, fostering collaboration but risking overlapping actions.
• On Citizens: Health care professionals benefit from protected autonomy in patient care, reducing fear of job repercussions for overriding potentially flawed AI (e.g., biased outputs). Patients may receive more personalized care, as human judgment is prioritized over automated suggestions.
• On International Relations: No direct impacts; the law focuses on domestic health care and employment practices.

Main Stakeholders Affected

• Health Care Professionals: Doctors, nurses, home aides, and insurance reviewers who use or override AI/CDSS; gain rights to independent decisions and protections from retaliation.
• Covered Entities: Hospitals, clinics, health plans, and employers in health care; must implement policies, training, committees, and face penalties for non-compliance.
• Government Entities: HHS and DOL for federal enforcement; state attorneys general and regulators for local actions; labor unions as representatives in committees and complaints.
• Patients: Indirectly affected through ensured human oversight in care decisions, potentially improving safety and equity.

Notable Legal, Constitutional, or Political Implications

• Legal: Establishes strong enforcement mechanisms, including private lawsuits with broad remedies and statutory damages, which could lead to increased litigation against employers. The non-preemption clause preserves state variations, allowing a patchwork of protections. Burdens of proof favor complainants, similar to anti-retaliation laws.
• Constitutional: Waives state sovereign immunity (under the 11th Amendment) for programs receiving federal funds, enabling suits against state entities and officials, which could expand federal oversight in public health care.
• Political: Promotes a balance between AI innovation in health care and worker safeguards, potentially influencing debates on technology regulation, labor rights, and health equity; emphasizes civil rights compliance in AI use, addressing biases without mandating AI bans.

This summary was generated by AI and may contain inaccuracies. Refer to the official source document for the authoritative text.

Sponsor

Sen. Markey, Edward J. [D-MA]

Cosponsors (2)

Sen. Blumenthal, Richard [D-CT], Sen. Smith, Tina [D-MN]

Recent Actions

• 2025-10-09: Read twice and referred to the Committee on Health, Education, Labor, and Pensions.
• 2025-10-09: Introduced in Senate

Bill Versions

• Right to Override Act — issued 2025-10-09 — PDF (28 pages)