AI in insurance claims and underwriting disputes

Insurers' use of AI raises questions about claims decisions, underwriting, pricing and governance. The NAIC's December 4, 2023 model bulletin addresses consumer-impacting AI systems, while New York's July 11, 2024 Circular Letter No. 7 focuses on underwriting and pricing. Medicare Advantage coverage decisions are also addressed in federal program guidance.

Claims and health-plan benefits

In its February 6, 2024 FAQ, CMS explained that Medicare Advantage organizations may use algorithms to assist coverage decisions while remaining responsible for compliance with the program's coverage rules. Medical-necessity determinations must consider the individual patient's circumstances. CMS specifically addressed predictions of post-acute-care length of stay and the need to consider the patient's condition and medical record before terminating coverage.

Estate of Gene B. Lokken v. UnitedHealth Group, Inc. provides a dated litigation example. Plaintiffs alleged that use of nH Predict displaced individualized decisions on post-acute care. On February 13, 2025, the District of Minnesota waived administrative exhaustion on futility grounds and allowed contract and implied-covenant claims to proceed, while dismissing other claims as preempted. The court recorded the parties' disagreement about use of the model and decided a motion to dismiss; the ruling addressed which pleaded claims could proceed. The dedicated case article covers subsequent proceedings.

On March 9, 2026, the court ordered production of relevant records about nH Predict's development and use, while excluding its underlying data, rules, source code and medical guidelines from the disputed request. Defendants filed a second amended answer denying liability on September 4, 2026.

Related legislative coverage includes the Protecting Patients from Automated Denials Act, whose sponsors announced its introduction on July 24, 2026. Their proposal would require physician review of AI-assisted Medicare Advantage prior-authorization denials and authorize federal audits of the relevant processes.

AI-assisted medical coding

In March and September 2026, the Blue Cross Blue Shield Association published analyses linking rising hospital diagnosis-coding intensity to higher commercial inpatient reimbursements. The March report examined claims from April 2022 through March 2025; the September report examined 2023–2025 trends and major bowel procedures. The association argued that AI-assisted documentation and coding could contribute to increases in billed complexity, using comparisons between recorded diagnoses and treatment patterns.

The American Hospital Association disputed the interpretation on October 5, 2026. It argued that changes in patient complexity and more complete documentation could explain coding trends, and that the analyses did not establish incorrect coding or identify which claims involved AI. These competing industry assessments concern providers’ billing and insurers’ reimbursement scrutiny, a separate issue from AI-assisted benefit denials.

Underwriting and pricing

New York's Circular Letter No. 7 sets supervisory expectations for insurers using AI and external consumer data in underwriting or pricing. It addresses actuarial validity, unfair discrimination, governance, documentation and third-party oversight. It also addresses consumer inquiries and complaints. These expectations are presented in the context of the New York laws and regulations cited in the letter.

Governance and evidence

The NAIC model bulletin calls for a written AI-systems program proportionate to an insurer's use of AI and the potential harm to consumers. It identifies governance, risk controls, testing and third-party arrangements as areas regulators may examine. Its model text leaves the issuing jurisdiction and local legal references for completion; application in a particular state requires checking that state's adopted instrument.

For a specific dispute, distinguish the challenged benefit or pricing decision, the contractual promise, the governing program or jurisdiction, and the role attributed to the system. The Lokken order illustrates why administrative exhaustion, preemption and the pleaded contract terms can determine which claims proceed before the merits of the alleged AI-assisted decision are resolved.

Singapore financial-sector supervision

On October 7, 2026, the Monetary Authority of Singapore (MAS) issued AI risk-management guidelines whose applicability listing includes direct insurers, reinsurers, captive insurers and insurance brokers. The supervisory expectations address board and senior-management oversight, AI inventories, risk assessment, lifecycle controls and third-party AI. Institutions remain accountable for AI used in their services. MAS sets phased implementation: sections 3–4 from October 7, 2027, and sections 5–6 by October 7, 2028. See Singapore AI Law for the broader financial-sector context.

Italy: AI Act market surveillance

Italy’s Legislative Decree No. 179/2026, dated October 7 and published October 8, 2026, entered into force on October 23, 2026. Article 5(1)(b) designates IVASS as the competent market-surveillance authority for high-risk AI systems used in insurance, reinsurance and insurance-product distribution governed by Italy’s Private Insurance Code. The allocation concerns supervision of AI used in insurance activities.

Related articles

• October 29 news: eHealth survey on human review of AI claim denials

• Insurance
• Insurance coverage disputes involving AI
• Protecting Patients from Automated Denials Act (119th Congress)
• New York AI Law
• Legislation and Regulation