When an Algorithm Acts as an Advisor: AI Liability Across Regulated Industries

On March 4, 2026, Nippon Life Insurance Co. of America sued OpenAI in the U.S. District Court for the Northern District of Illinois, alleging that ChatGPT practiced law without a license. The insurer claims that a disability benefits claimant used ChatGPT to generate legal arguments, draft court filings, and ultimately pursue litigation that had no procedural basis. The complaint describes a situation in which an AI tool encouraged a user to breach a settlement agreement, fire her attorney, and file 44 post-settlement motions and related filings, including at least one fabricated case citation. The complaint advances three causes of action, unauthorized practice of law, tortious interference with contract, and abuse of process, and seeks $300,000 in compensatory damages plus $10 million in punitive damages.

The case is among the first to name an AI company as a defendant for unauthorized practice of law. It is also filed by an insurer, not a law firm or bar association, which signals that the downstream effects of unregulated AI advice are spreading well beyond the legal profession.

The Legal Sector Is Already Responding

This lawsuit follows a series of legal developments in 2025 and early 2026. In February 2026, a federal judge in the Southern District of New York ruled that a defendant's communications with a generative AI chatbot are not protected by attorney-client privilege. That ruling established that public AI platforms are functionally third parties, and disclosing information to them may waive privilege. Courts have also encountered fabricated case citations generated by AI tools, and New York's Senate has advanced a bill that would create a private right of action against chatbot operators who provide responses that would constitute unauthorized practice of a licensed profession.

The regulatory pattern is clear. AI tools that produce outputs resembling professional advice will be held to professional standards, or their operators will face liability for failing to prevent that outcome.

Insurance: The Sector With the Most to Lose and to Learn

The Nippon case is not just a legal story. It is an insurance story. The plaintiff is a life insurer that absorbed the operational and reputational cost of AI-generated litigation. That makes this lawsuit a direct example of the risks insurance companies face across multiple fronts as AI adoption accelerates.

Consider the exposure points. On the claims side, insurers are increasingly deploying AI for intake, adjudication, and fraud detection. A Sedgwick report published in March 2026 estimates that between 58% and 82% of carriers already use AI in their operations, but only 12% report mature capabilities, and 75% of claims professionals say AI still requires human oversight. Fragmentation across vendors and systems is leading to data and decision-quality inconsistencies. Insurers using AI to deny claims have already been sued: Cigna faced a class action over its PXDX algorithm, which automates claims denials without physician review, and UnitedHealthcare was sued over its nH Predict model, where a federal judge has since allowed broad discovery into the system's role in Medicare Advantage claim denials.

On the regulatory side, the NAIC's AI Systems Evaluation Tool moved from development into pilot examinations in early 2026. Twenty-four states have adopted the NAIC's model bulletin on insurers' use of AI. New York's Circular Letter No. 7 sets specific standards for AI in underwriting and pricing. The EU AI Act classifies insurance risk assessment and pricing AI as high-risk, and although the Digital Omnibus enacted in July 2026 deferred the standalone high-risk obligations to December 2027, the transparency obligations took effect in August 2026 and the compliance clock for carriers with EU operations is running.

The litigation landscape is expanding as well. Chatbot-related wiretap lawsuits grew from 2 matters in 2021 to 30 in 2025, and healthcare insurers and life insurance companies were among the defendants. Industry analysts expect AI-related insurance claims to increase substantially through 2026 and beyond, spanning E&O, cyber liability, media liability, and D&O coverage lines.

Finance and Other Regulated Sectors

The same dynamics apply in financial services. AI tools that generate investment guidance, tax advice, or compliance interpretations face the same question: at what point does an AI output cross the line from general information to regulated professional advice? SEC requirements around suitability and fiduciary duty do not have carve-outs for AI-generated recommendations. A chatbot that steers a retail investor toward a strategy without disclosure of risk, conflicts of interest, or licensing is operating in the same territory as the OpenAI and Nippon scenario.

Healthcare faces parallel risks. Stanford researchers published findings in January 2026 showing that both older algorithms and AI systems in health insurance make frequent errors in claims processing, and that confusing denial letters generated by automated systems discourage appeals. Two-thirds of U.S. adults report little trust that AI will be used responsibly in healthcare, and health insurers rank among the least trusted sectors.

Building Toward a Risk Framework

AI risk is no longer speculative. Carriers, financial institutions, and any organization deploying AI in regulated processes need a structured approach to identifying, categorizing, and mitigating these risks before they materialize as litigation, regulatory action, or operational failure. The question raised by the Nippon case, who bears responsibility when an algorithm acts as an advisor, will be answered either by deliberate governance now or by courts and regulators later.

I have published a comprehensive AI Adoption Risk Framework for the insurance industry through Arctyra, covering the risk taxonomy, control framework, regulatory compliance mapping, and governance maturity model that this landscape demands. The executive summary is available at arctyra.ai.

Sources

ABA Journal, "OpenAI sued for practicing law without a license," March 2026. Norton Rose Fulbright, "Complaint accuses OpenAI of practicing law without a license" (Nippon Life Ins. Co. of Am. v. OpenAI Fdn., No. 1:26-cv-02448, N.D. Ill.), April 2026. Goodwin Law, "AI Chatbots, Privilege, and Pitfalls" (Heppner ruling, SDNY), March 2026. Holland & Knight, "New York Bill Would Create Liability for Chatbot Proprietors" (SB 7263), March 2026. Sedgwick, "Future-ready property claims: Leveraging technology and AI," March 2026. Baker Botts, "AI Chatbot Regulation: 78 State Bills, 58 Lawsuits," 2026. Fenwick, "Tracking the Evolution of AI Insurance Regulation," December 2025. Stanford Report, "AI-driven insurance decisions raise concerns about human oversight," January 2026. Olshan Law, "Expect an Uptick in AI Insurance Related Claims in 2026." IAPP, "How AI liability risks are challenging the insurance landscape."