States Eye AI Oversight
What happened
- State legislatures are considering oversight of AI used in health-insurance decisions like prior authorization. - Legal proposals focus on transparency and governance of insurer algorithms for utilization review. - New rules could force clearer disclosure about why denials happen, changing payer-practice interactions. (natlawreview.com)
Why it matters
State lawmakers are moving to limit how health insurers use artificial intelligence in prior authorization, the gatekeeping process that decides whether care gets approved. (natlawreview.com) The current push centers on utilization review, the insurer process for judging whether a test, drug, or procedure is medically necessary before payment. A National Law Review survey published April 22 said Pennsylvania and New Hampshire are among the states weighing 2026 bills after Arizona, Maryland, Nebraska, and Texas enacted measures in 2025. (natlawreview.com; ama-assn.org) The bills generally do not ban software outright. They let insurers use algorithms as assistive tools, but they block plans from letting automated systems replace a clinician’s judgment in adverse decisions such as denials, delays, or service cuts. (natlawreview.com; ama-assn.org) Pennsylvania’s House Bill 1925 and Senate Bill 1113 would require adverse benefit determinations to be based on a patient’s clinical history, the requesting provider’s stated circumstances, and other information in the medical record. The bills would also require annual artificial-intelligence compliance filings and disclosure of artificial-intelligence use to providers and members. (natlawreview.com) New Hampshire’s House Bill 1406 would allow artificial intelligence in utilization review as an assistive tool, require insurers to keep written records on its use, and require adverse determinations to be made by a qualified health care provider. The proposal would also require written notices explaining the basis for an adverse decision and whether artificial intelligence was used in the review. (natlawreview.com) The state laws already on the books show the pattern. Nebraska’s 2025 law says an artificial-intelligence algorithm cannot be the sole basis for denying, delaying, or modifying care on medical-necessity grounds, and it requires plans to disclose algorithm use to regulators, providers, enrollees, and on public websites. (nebraskalegislature.gov; law.justia.com) Texas went further in Senate Bill 815, which took effect September 1, 2025. The law bars a utilization review agent from using an automated decision system to make, wholly or partly, an adverse determination, while still allowing algorithms for administrative support and fraud detection; it also gives the insurance commissioner audit authority. (legiscan.com; capitol.texas.gov) Maryland’s House Bill 820, approved in May 2025, requires carriers, pharmacy benefit managers, and private review agents to make sure software tools rely on a patient’s individual medical history and clinical circumstances rather than group datasets. The law also requires quarterly reporting to the insurance commissioner on whether artificial intelligence or other software tools were used in adverse decisions. (mgaleg.maryland.gov; ama-assn.org) Washington added another version of the rule on March 24, when Governor Bob Ferguson signed Senate Bill 5395. The law bars private insurers and the state employee health program from using artificial intelligence as the only means to deny, delay, or modify care, and denial notices must identify the reviewing provider’s credentials, board certifications, and specialty. (senatedemocrats.wa.gov) The backdrop is a wider surge in state artificial-intelligence lawmaking. The National Conference of State Legislatures says its database, updated April 1, tracks enacted and pending bills across health care, oversight, discrimination, and other topics beginning in 2025, while the American Medical Association said more than 250 state bills on artificial intelligence in health care had been introduced by November 2025. (ncsl.org; ama-assn.org) For insurers, the immediate compliance shift is less about abandoning software than documenting it, disclosing it, and preserving a human reviewer’s independent judgment. For patients and doctors, the practical change is that denial notices in more states may have to say more clearly who made the decision, what records were reviewed, and whether a machine helped shape the answer. (natlawreview.com; senatedemocrats.wa.gov)
Key numbers
- A National Law Review survey published April 22 said Pennsylvania and New Hampshire are among the states weighing 2026 bills after Arizona, Maryland, Nebraska, and Texas enacted measures in 2025.
- (nebraskalegislature.gov; law.justia.com) Texas went further in Senate Bill 815, which took effect September 1, 2025.
- (mgaleg.maryland.gov; ama-assn.org) Washington added another version of the rule on March 24, when Governor Bob Ferguson signed Senate Bill 5395.
What happens next
- They let insurers use algorithms as assistive tools, but they block plans from letting automated systems replace a clinician’s judgment in adverse decisions such as denials, delays, or service cuts.
- For patients and doctors, the practical change is that denial notices in more states may have to say more clearly who made the decision, what records were reviewed, and whether a machine helped shape the answer.
- New rules could force clearer disclosure about why denials happen, changing payer-practice interactions.
Quick answers
What happened in States Eye AI Oversight?
State legislatures are considering oversight of AI used in health-insurance decisions like prior authorization. Legal proposals focus on transparency and governance of insurer algorithms for utilization review. New rules could force clearer disclosure about why denials happen, changing payer-practice interactions. (natlawreview.com)
Why does States Eye AI Oversight matter?
State lawmakers are moving to limit how health insurers use artificial intelligence in prior authorization, the gatekeeping process that decides whether care gets approved. (natlawreview.com) The current push centers on utilization review, the insurer process for judging whether a test, drug, or procedure is medically necessary before payment. A National Law Review survey published April 22 said Pennsylvania and New Hampshire are among the states weighing 2026 bills after Arizona, Maryland, Nebraska, and Texas enacted measures in 2025. (natlawreview.com; ama-assn.org) The bills generally do not ban software outright. They let insurers use algorithms as assistive tools, but they block plans from letting automated systems replace a clinician’s judgment in adverse decisions such as denials, delays, or service cuts. (natlawreview.com; ama-assn.org) Pennsylvania’s House Bill 1925 and Senate Bill 1113 would require adverse benefit determinations to be based on a patient’s clinical history, the requesting provider’s stated circumstances, and other information in the medical record. The bills would also require annual artificial-intelligence compliance filings and disclosure of artificial-intelligence use to providers and members. (natlawreview.com) New Hampshire’s House Bill 1406 would allow artificial intelligence in utilization review as an assistive tool, require insurers to keep written records on its use, and require adverse determinations to be made by a qualified health care provider. The proposal would also require written notices explaining the basis for an adverse decision and whether artificial intelligence was used in the review. (natlawreview.com) The state laws already on the books show the pattern. Nebraska’s 2025 law says an artificial-intelligence algorithm cannot be the sole basis for denying, delaying, or modifying care on medical-necessity grounds, and it requires plans to disclose algorithm use to regulators, providers, enrollees, and on public websites. (nebraskalegislature.gov; law.justia.com) Texas went further in Senate Bill 815, which took effect September 1, 2025. The law bars a utilization review agent from using an automated decision system to make, wholly or partly, an adverse determination, while still allowing algorithms for administrative support and fraud detection; it also gives the insurance commissioner audit authority. (legiscan.com; capitol.texas.gov) Maryland’s House Bill 820, approved in May 2025, requires carriers, pharmacy benefit managers, and private review agents to make sure software tools rely on a patient’s individual medical history and clinical circumstances rather than group datasets. The law also requires quarterly reporting to the insurance commissioner on whether artificial intelligence or other software tools were used in adverse decisions. (mgaleg.maryland.gov; ama-assn.org) Washington added another version of the rule on March 24, when Governor Bob Ferguson signed Senate Bill 5395. The law bars private insurers and the state employee health program from using artificial intelligence as the only means to deny, delay, or modify care, and denial notices must identify the reviewing provider’s credentials, board certifications, and specialty. (senatedemocrats.wa.gov) The backdrop is a wider surge in state artificial-intelligence lawmaking. The National Conference of State Legislatures says its database, updated April 1, tracks enacted and pending bills across health care, oversight, discrimination, and other topics beginning in 2025, while the American Medical Association said more than 250 state bills on artificial intelligence in health care had been introduced by November 2025. (ncsl.org; ama-assn.org) For insurers, the immediate compliance shift is less about abandoning software than documenting it, disclosing it, and preserving a human reviewer’s independent judgment. For patients and doctors, the practical change is that denial notices in more states may have to say more clearly who made the decision, what records were reviewed, and whether a machine helped shape the answer. (natlawreview.com; senatedemocrats.wa.gov)
