Life Insurance Client Communication. The ABA and certain states treat communicating AI use to a client as circumstancedependent, only required when inputting confidential information or influencing a significant decision. Kentucky and North Carolina specify that routine AI use, such as for research, need not be communicated. Illinois and Pennsylvania, however, advise disclosure of any AI use. Fees. The ABA and nearly all states weighing in agree that lawyers may not charge hourly fees for time saved by AI. But certain states like Maryland provide more flexible guidance, stating that if AI substantially reduces the time to perform a task, that may trigger a reconsideration of the amount to charge for those legal services. Competence. There is a consensus that attorneys must understand the risks and benefits of the AI tools they use. But the ABA and certain states go further, suggesting that competence may require becoming familiar with AI tools. The ABA Opinion notes that “it is conceivable that lawyers will eventually have to use [AI] to competently complete certain tasks for clients,” much like email or word processing. These interpretations highlight the importance of clear communication and agreements between in-house and outside counsel on expectations for AI use. n Et Tu, Judge? “There are several federal judges who are beta testing Generative AI.” – survey respondent A recent article in the Sedona Conference Journal reports findings from a randomized survey of federal judges on AI use, conducted by the New York City Bar Association Presidential Task Force and Northwestern University. As of December 2025, over 60% of judges reported using at least one AI tool in their judicial work; 22.4% used AI daily or weekly; and one-third permitted or encouraged AI use in their chambers. Judges primarily employed AI for legal research and document review. Judicial adoption may accelerate, partly motivated by a surge in filings from litigants using generative AI. In February, Los Angeles County launched a pilot program providing six judges with access to Learned Hand, a judicial-focused AI software designed to distill voluminous filings, assist with judicial reasoning, and draft preliminary orders. The Michigan Supreme Court has used the software since 2025 to help review applications for leave to appeal, and eight other court systems have reportedly contracted with the platform. n “A Modest Proposal” for AI in Contract Interpretation Snell v. United Specialty Insurance Company, No. 22-12581 (11th Cir. May 28, 2024). “My only proposal—and, again, I think it’s a pretty modest one—is that we consider whether LLMs might provide additional datapoints to be used alongside dictionaries, canons, and syntactical context in the assessment of terms’ ordinary meaning.” — Hon. Kevin C. Newsom, concurring Of course, some judges will be faster to adopt AI than others. It’s worth mentioning—and reading— a concurring opinion written two years ago by Judge Newsom of the Eleventh Circuit. The coverage dispute involved a question about whether “landscaping” included installing an in-ground trampoline. After “languishing in definitional purgatory,” Judge Newsom explored how AI large language models (LLMs) could have some advantages over traditional dictionaries: the broad range of everyday language usage they are trained on, the ability to set words in context, and a process that is sometimes more accessible, democratic, and transparent than the human judgment involved in assembling and selecting a dictionary definition. In short, they predict “how ordinary people use words and phrases in context,” which in some circumstances could be a helpful interpretation tool. All About AI 5 4 All About AI Five Things to Know About AI 1. A I in the Product Life Cycle. The use of AI is not limited to insurance underwriting or claims decisions—it is also used in development and design, marketing, utilization management, pricing, customer service functions, and fraud detection. 2. E rrors Scale Quickly. For example, straightforward claims decisions can be made in minutes as opposed to days. With this increased speed and cost-saving, though, risk can be magnified because errors scale just as quickly. 3. R egulation of AI in the Insurance Context Is Fragmented and State-Driven. Some states with formal laws regulating AI often require (1) that utilization review emphasize initialized decisions, focusing, for example, on specific clinical data from a policyholder’s medical history rather than generalized data; (2) insurers to disclose how AI tools are being used; and (3) that AI systems are applied fairly and without discrimination. Other states, like Texas, focus on preventing intentional discrimination only. 4. Litigation Trends ▪ Proxy discrimination – Several property insurance cases have been filed alleging algorithmic bias in AI claims handling tools. The legal theory is that although the algorithms don’t use explicit inputs like race or disability status, they use proxy variables (like voices, names, or credit scores) that overlap with protected classes, resulting in delayed payments, heightened scrutiny, and disparate treatment of elderly, disabled, or Black policyholders. ▪ Lack of meaningful human oversight – Courts have allowed breach of fiduciary duty and unfair competition law claims to proceed on allegations that an insurer delegated health insurance claims handling to an algorithm to determine whether care was a medical necessity, subverting required review by medical professionals. The cases implicate the opacity of AI algorithms, which can make it hard to know why a particular determination was made and in turn to challenge the determination. ▪ No actuarial justification – Other cases allege that the use of nontraditional data sources (including consumer data) was not actuarially justified or created unfair discrimination. 5. N AIC at Odds with the White House. The NAIC stated concerns about the sweeping Executive Order directing a national policy framework for AI regulation, asking that the administration instead affirm state regulation of AI in the business of insurance. From the NAIC’s perspective, the Executive Order could be read to potentially prevent regulators from addressing risks in ratemaking setting, underwriting, and claims handling. n
RkJQdWJsaXNoZXIy Njk5MDg5