Source: North Carolina Criminal Law Blog

Generative AI tools now produce legal summaries and draft pleadings, expanding access to legal information while raising risks of frivolous or unsupported filings and potential exposure of litigant information when open-source tools replace counsel. GenAI can also alter or fabricate evidence—such as modifying recordings or generating false material—creating authenticity and reliability challenges that trial courts must address. Responses across courts vary: some ban GenAI drafting, some mandate disclosure and attorney verification, and others decline mandatory disclosure but stress adherence to ethical rules. A Texas court requires certification that AI-created content was verified by a Texas-licensed attorney. The Illinois Supreme Court’s policy permits AI use without mandatory disclosure but underscores that the Rules of Professional Conduct and Code of Judicial Conduct fully apply and that users remain accountable for final filings.

North Carolina’s Supreme Court has not issued a statewide AI rule; the North Carolina State Bar issued a 2024 Formal Ethics Opinion on attorney responsibility when using AI. Superior Court District 25 (Cabarrus County) adopted Revised Administrative Order 25-09 (December 8, 2025, replacing the July 23, 2024 order), promulgated by Senior Resident Superior Court Judge Martin McGee, which distinguishes AI use for advocacy from AI’s use to generate or alter evidence. The order requires disclosure of AI-generated or altered evidence—generally 90 days before trial—while not requiring disclosure for advocacy drafting, and includes a chart and Quick Reference Guide. Judge McGee, a RAILS fellow, is developing a model administrative order. The National Center for State Courts and Thomson Reuters Institute created the AI Policy Consortium for Law & Courts and produced resources on AI-generated evidence, hallucinations, and guidance for court users. Federal Judge Xavier Rodriguez reported using GenAI to summarize testimony and draft findings, cautioned against substituting judicial decision-making with AI due to automation and confirmation bias, found GenAI outputs overbroad and superficial, and advised judges to understand case facts, tool limitations, and to rigorously evaluate AI results.