In Estate of Gene B. Lokken, et al. v. UnitedHealth Group, Inc., Case No. 0:23-cv-03514 (D. Minn.), a putative class has brought suit against UnitedHealth Group, Inc., naviHealth, Inc., and United Health Group (collectively “UHC”), alleging that UHC used an artificial intelligence program called nH Predict to deny medical care coverage. The putative class is made up of Medicare Advantage customers of UHC who sought coverage for post-acute care and were eventually denied coverage.

In their complaint, the plaintiffs allege they purchased Medicare Advantage health insurance plans from UHC that contained a provision that “[UHC’s] Clinical Services Staff and Physicians make decisions on the health care services you receive based on the appropriateness of care and service and existence of coverage.” However, the plaintiffs alleged that despite this written requirement, UHC used nH Predict, in lieu of physicians, to make coverage determinations. The plaintiffs further alleged that the AI model was used to determine the amount of coverage an individual patient needed, regardless of recommendations made by their own treating physician. The plaintiffs alleged that UHC knew of nH Predict’s inaccuracies because over 90% of claim denials were reversed on appeal, and over 80% of preauthorization denials were reversed. UHC denies any use of nH Predict. The class alleged that UHC’s AI-generated denials of treatment recommended by their treating physicians resulted in worsening injuries and, in some cases, death.

Initially, UHC moved to dismiss the class action, arguing that the class’s claims are subject to the exhaustion of administrative remedies and/or that the class’s claims are preempted by the Medicare Act. The motion was granted in part and denied in part on the basis that certain claims were preempted by federal statute. However, the putative class’s breach of contract and breach of the implied covenant of good faith and fair dealing claims were allowed to proceed to discovery, with the district judge noting that these claims “effectively arise out of [UHC’s] evidence of coverage documents because the question [is] whether UHC complied with its statement that claim decisions would be made by ‘clinical services staff’ and ‘physicians’ when it allegedly used artificial intelligence.” Est. of Lokken v. UnitedHealth Grp., Inc., 766 F. Supp. 3d 835, 848 (D. Minn. 2025).

Several discovery disputes have since arisen, with a magistrate judge recently issuing an order compelling UHC to produce documents and information related to the development and use of nH Predict, governmental investigations related to UHC’s use of nH Predict, or AI tools in relation to the assessment and adjudication of post-acute care claims, and UHC’s oversight of AI use. See Est. of Lokken v. UnitedHealth Grp., Inc., 2026 WL 658883 (D.Minn. 2026). The court found that the class was entitled to the discovery of documents regarding how nH Predict works, its development goals and anticipated benefit, and whether it was designed to supplant physician decision-making. Despite UHC’s contentions that the discovery was sought based on a “faulty premise” that nH Predict was used to make coverage decisions, the court held that whether the AI tool made coverage decisions was a “merits-based argument” that did not foreclose the requested discovery.

Lokken is just one example of cases involving the use of artificial intelligence in claims decision-making by health insurers. See, e.g., Estate of Barrows v. Humana, Inc., 3:23-cv-654-RGJ (W.D. Ky); Kisting-Leung v. Cigna Corp., 2:23-cv-01477-DAD-CSK (E.D. Cal.).

Although Lokken arises out of the context of medical insurance, there may be risks for insurers utilizing AI in handling claims under all types of insurance policies. A recent study found that 70% of automobile, homeowners, and health insurers responding to the surveys are currently using, planning to use, or exploring the use of AI. While 58% of life insurer respondents report current or expected future use. See Artificial Intelligence and Insurance Regulation, Journal of Insurance Regulation, Vol. 45, Number 3, National Association of Insurance Commissioners, Innovation Cybersecurity and Technology Committee, May 5, 2026. At present, AI use for these insurers has been primarily limited to underwriting, rating, and pricing, with plans to develop technology to automate approval and denial decisions and process renewals and reinstatements. See id. at pp. 9-10.

Lokken may provide a blueprint of the types of actions that may be brought and how courts will deal with claims against insurers in first-party actions, bad faith claims, and coverage matters involving an insurer’s use of AI. In addition, with the federal government and numerous states passing statutes regulating the use of AI, including by insurers, insurers may face statutory claims brought by individuals or states. See, e.g., Texas Senate Bill 815, 2025 (prohibiting the use of an automated system (an algorithm) from making, wholly or partly, an adverse determination). Insurers should be aware of the potential for these types of cases and anticipate insureds requesting information regarding an insurer’s AI use through discovery in many types of coverage cases.

Moving forward, insurers should anticipate the potential for increased litigation risk and potential regulatory attention related to AI use in claims handling and coverage determinations. They should carefully evaluate whether their practices align with policy and other representations, as well as insureds’ expectations regarding fairness and accuracy.

About Kari Shane

Kari Shane is an associate in Tressler’s Insurance Practice Group in the Chicago Office. Kari advises insurers on policy matters, including professional liability, fidelity bonds, directors’ and officers’ liability, commercial general liability, employment practices liability, and first-party property claims. Additionally, she handles insurance coverage analysis disputes and litigation. Click here to read Kari’s full attorney biography.