As artificial intelligence (AI) has become an increasingly common presence in daily life, policymakers have been considering ways to ensure that the use of AI is not replacing human expertise and decision-making on complex topics. One of the areas of concern for both physicians and policymakers has been AI’s use in health insurance, particularly the ways in which insurance companies have begun utilizing AI to assess benefit coverage decisions.
How AI is Being Used in Health Insurance
Health Insurers
In 2025, the National Association of Insurance Commissioners published its Artificial Intelligence and Machine Learning (AI/ML) Survey Report, representing feedback from 93 insurance companies in 16 states. Of the companies surveyed, 84 percent indicated they use AI/ML across product lines. Companies selling individual major medical health insurance reported currently using or exploring the use of AI/ML for:
- Utilization management practices (71.0 percent);
- Disease management programs (61.0 percent);
- Prior authorization for approval processes (68.0 percent);
- Claims fraud detection (50.0 percent);
- Medical provider fraud detection (51.0 percent);
- Sales and marketing solutions (45.0 percent); and
- Denying prior authorizations (12.0 percent).
One of the key ways that insurers have begun using AI in their work is through prior authorization. Prior authorization requires health care providers to obtain approval from a patient’s health insurer before providing the prescribed item, service, or prescription.
A February 2025 study by the Medicaid and Children’s Health Insurance Plan (CHIP) Payment and Access Commission (MACPAC) notes that some studies suggest prior authorization can “reduce health care costs without negatively impacting care quality,” and it has been successfully used to reduce overutilization of some items, redirect care to less expensive treatments, and help ensure care aligns with accepted clinical standards by not covering experimental treatments or non-approved uses of medications. However, prior authorization can also cause necessary care to be delayed or denied, declines in health, and increased administrative burden and cost.
Physician Concerns
A survey conducted in 2024 by the American Medical Association found that 61.0 percent of physicians surveyed reported concern that the use of AI by health plans is increasing prior authorization denials. With these increased denials, surveyed physicians also reported poor clinical outcomes for patients (94 percent), delayed care (93 percent), and increased administrative burden for physicians and staff.
Class Action Lawsuit
A federal class action lawsuit filed in Minnesota against UnitedHealthcare in 2023 claims the company is “wrongfully denying elderly patients care owed to them under Medicare Advantage Plans by overriding their treating physicians’ determinations as to medically necessary care based on an AI model that Defendants know has a 90 percent error rate” (Estate of Gene B. Lokken et al. v. UnitedHealth Group, Inc.). In February 2025, the U.S. District Court for the District of Minnesota denied, in part, a request by UnitedHealthcare to dismiss the lawsuit. Portions of the lawsuit regarding breach of contract and breach of the implied covenant of faith and fair dealing were allowed to proceed. The Court also waived the requirement that plaintiffs must exhaust the Medicare appeals process prior to filing a lawsuit due to the potential for irreparable harm to many patients.
Policy Responses
Recent State Actions
Colorado. Colorado became the first state to implement regulations on AI and insurance on November 14, 2023. The regulations require life insurance companies to report how they review AI models and use External Consumer Data and Information Sources (ECDIS), which may include data like credit scores, social media habits, purchasing habits, home ownership, educational attainment, licensures, court records, and other information to supplement or supplant traditional underwriting factors. Life insurance companies are also required to develop a governance and risk management framework. Similar regulations are set to take effect for health and auto insurers in Colorado on October 15, 2025.
California. In 2024, the California Legislature passed SB 1120, the Physicians Make Decisions Act (Act), to restrict health insurers and disability insurers from using AI, algorithms, and similar tools as the sole means to deny, delay, or modify care based on medical necessity. The Act, enacted on January 1, 2025, provides that final medical necessity determinations may be made only by a licensed physician or licensed health care provider who is competent in the specific clinical issues and services requested by the provider.
The bill applies to prospective, retroactive, and concurrent utilization functions and creates requirements for AI, algorithms, and other tools used in health care. Under the bill, these tools must:
- Base decisions regarding medical necessity on the enrollee’s medical or clinical history and circumstances;
- Be applied fairly and equitably;
- Be available to inspection for audit and compliance reviews by specified state agencies; and
- Be reviewed periodically to assess outcomes and ensure accuracy, reliability, and compliance with the Act.
2025 State Legislative Actions
During the 2025 Legislative Session, four states passed bills to prohibit payors from using AI to deny medical necessity or prior authorization determinations.
Arizona. HB 2175 requires that health care providers independently review claims and prior authorization requests prior to an insurer denial. The sole use of any other source to deny a claim or prior authorization is prohibited.
Maryland. HB 820 requires carriers to ensure that if an AI tool is used for utilization review, it must include medical history, individual circumstances, and other clinical information in its determination. Such AI tools must be open for inspection and audit by the state, and patient data may not be used for any purpose beyond the intended utilization review.
Nebraska. LB 77 prohibits AI output from being the sole basis of evaluating medical necessity in order to deny, delay, or modify health care services. The use of AI in utilization review must be disclosed to health care providers and enrollees and communicated clearly on the payor’s public website.
Texas. SB 815 prohibits the use of AI to make adverse determinations regarding medical necessity of services. Payors may only use AI for administrative support or fraud detection. The bill also requires any AI tools in use by payors to be subject to inspection by the Commissioner of Insurance.
Federal Actions
The Centers for Medicare and Medicaid Services (CMS) issued a Final Rule, effective January 2024, stating that Medicare Advantage (MA) plans may use AI, algorithms, and related technologies to make coverage determinations, but medical necessity determinations must be “based on the circumstances of each specific individual” and determinations “must be reviewed by a physician or other appropriate health care professional with expertise in the field of medicine or health care that is appropriate for the services at issue.”
In November 2024, CMS released a Fact Sheet discussing proposed changes to the MA Program for Contract Year 2026, including requiring MA plans to “ensure services are provided equitably, irrespective of delivery method or origin, whether from humans or automated systems.” When the Final Rule for the 2026 MA Program was released in April 2025, the rule did not include these proposed guardrails on AI, but CMS noted broad interest in the regulation of AI and stated the agency would “continue to consider the extent to which it may be appropriate to engage in future rulemaking in this area.”
By Leighann Thone and Iraida Orr
See Health and Social Services for more.