Artificial intelligence has been quietly reshaping the legal profession for the past few years, and now state courts across America are catching up with formal rules to govern how lawyers can and cannot use it. Rhode Island became the latest state to act this week, with its highest court amending professional conduct rules and releasing detailed guidance on generative use in legal practice. The move signals that the era of informal AI adoption in law is over. The guardrails are going up, AI Rules for Lawyers and they are going up fast.

How AI Entered the Legal Profession Without a Rulebook

When large language models became widely accessible in 2022 and 2023, law firms of every size began experimenting almost immediately. The appeal was obvious. AI tools could draft briefs, summarise case law, generate contract language, and cut research time dramatically. Small firms saw an opportunity to compete with larger ones. Solo practitioners saw a lifeline. The technology spread through the legal industry with remarkable speed and very little oversight.

The problem was that courts, bar associations, and disciplinary bodies had no specific rules in place to govern any of it. Lawyers were using powerful generative AI tools in live legal proceedings without any formal guidance on verification, disclosure, billing, or professional responsibility. Most existing conduct rules had been written long before this technology existed and offered no clear answers to the questions practitioners were suddenly facing every day.

AI Hallucinations Started Costing Lawyers Their Reputations

The consequences of unregulated AI use in law became public and embarrassing very quickly. Several high-profile cases emerged in which attorneys submitted court filings containing citations to cases that simply did not exist. These fabricated references, known as AI hallucinations, were generated by language models that confidently produced plausible-sounding but entirely fictional legal precedents. Judges were not amused.

Courts across the country began sanctioning attorneys who had submitted AI-assisted work without properly verifying its contents. Some lawyers faced fines. Others faced formal disciplinary proceedings. A few cases drew national media attention, putting the entire legal profession on notice that AI was not a magic shortcut but a tool requiring serious human oversight. The reputational damage to those attorneys was significant, and the message to the broader profession was unmistakable.

Rhode Island Acts as a Wave of State Regulations Builds

The Rhode Island Supreme Court moved on two fronts simultaneously this week. First, it amended the state's professional conduct rules to formally require lawyers licensed in Rhode Island to stay current on changes in the law, specifically including the benefits and risks associated with existing and developing technology. This is not optional continuing education. It is a binding professional obligation with disciplinary implications if ignored.

The court also issued detailed advisory guidance focused specifically on generative AI. The guidance is direct and practical. Lawyers must independently review and verify any AI-generated work before submitting it, precisely because these tools can hallucinate legal citations or mischaracterise legal precedent in ways that are difficult to detect without careful human review. The guidelines acknowledge that this is a fast-moving space, noting they will be subject to ongoing revision as the generative AI landscape continues to evolve. Rhode Island is not pretending it has all the answers. It is establishing a floor and promising to keep building.

Billing Rules and Judicial Independence Take Centre Stage

Two specific areas of the Rhode Island guidance deserve particular attention because they address questions that have been generating debate inside law firms for months. On legal fees, the court was clear. Lawyers are prohibited from billing clients for time saved through AI efficiencies. If a task that once took four hours now takes 45 minutes because of AI assistance, the client does not pay for four hours. Routine AI costs must generally be treated the way firms treat overhead expenses like rent, productivity software, and malpractice insurance.

The guidance also extends its reach to judges, which is unusual and significant. The Rhode Island Supreme Court explicitly stated that the use of artificial intelligence technologies must not undermine or replace the independent decision-making of judges. This is the court drawing a very clear line. AI can assist. It cannot decide. The human judgment of a trained jurist must remain at the centre of every legal determination, regardless of what tools are used to inform that judgment.

Other States Are Moving in the Same Direction

Rhode Island is not acting in isolation. It is part of a clear national pattern that has been building momentum throughout 2025 and into 2026. The Florida Supreme Court moved in May, requiring lawyers to certify that every case they reference in a filing actually exists and is accurately cited. That certification requirement puts personal professional responsibility squarely on the attorney, not the AI tool they used.

New York followed with its own rule, effective June 1, requiring attorneys to ensure that filings contain no fabricated or fictitious cases, statutes, or other material. New York's rule is particularly significant given the sheer volume of federal and state litigation that flows through its courts every year. Taken together, Florida, New York, and Rhode Island represent a significant portion of the American legal market, and their coordinated movement toward AI regulation is sending a message to every other jurisdiction still sitting on the sidelines.