The first wave of litigation challenging insurers’ use of artificial intelligence (“AI”) in coverage determinations has arrived. These early decisions suggest that courts willingly scrutinize AI’s role in claim handling and are willing to conclude that the reasonable use of an AI platform mustn’t be criticized. We’ve briefly addressed three (3) recent federal decisions – all of which were well-reasoned – in support of our narrative here.
In Kisting-Leung v. Cigna Corp., plaintiffs alleged that Cigna improperly relied on its PxDx AI tool to deny claims without adequate disclosure. The record there, however, revealed that claim denial letters identified the involvement of an algorithm. The U.S. District Court for the Eastern District of California found no evidence that any alleged injury, or denial of benefits, was traceable to any unlawful conduct by Cigna, even assuming AI played a role. Similarly, in Trzeciak v. Allstate Property & Casualty Insurance Co., plaintiffs alleged that Allstate used a form of AI to overcharge premiums, while also failing to even disclose the practice. The U.S. District Court for the Eastern District of Michigan dismissed the Complaint for lack of standing, emphasizing that the Allstate had no independent duty to disclose its use of AI when policy documents already referenced proprietary models and clearly described the premium charged. By contrast, in Estate of Lokken v. UnitedHealth Group Inc., plaintiffs challenged UnitedHealth’s use of the nH Predict tool to deny coverage and requested further discovery into its use. The U.S. District Court for the District of Minnesota allowed expansive discovery into the system’s design, implementation, and real-world application, reasoning that such information could illuminate changes in claims-handling practices tied to AI adoption.
These opinions suggest that courts are pursuing a practical and readily identifiable standard for evaluating the use of AI in coverage determinations. Currently, the mere use of AI, standing alone, is not independently actionable. However, where plaintiffs plausibly allege that AI materially contributed to a wrongful denial of coverage or otherwise caused a cognizable harm, courts are likely to permit such claims to proceed and all discovery into the role the technology played in the decision-making process.