The University of Chicago Law School, Berkeley Law and Columbia Law School are among a growing group of U.S. law schools rewriting their rules on artificial intelligence as legal educators try to preserve independent legal reasoning while preparing students for AI-heavy professional practice.
At least a dozen U.S. law schools introduced new or revised AI policies over the summer, according to Reuters, ranging from restrictions on classroom technology to compulsory instruction in legal technology. The changes reflect a shift from treating generative AI mainly as an academic-integrity problem towards deciding when students should work without it and when they should be trained to use it.
The University of Chicago has adopted one of the more restrictive approaches for its required first-year core curriculum. During the 2026-27 academic year, laptops, tablets and phones are prohibited in those classes subject to limited exceptions, including disability accommodations and instructor-authorised activities. Examinations will also take place without access to the internet, electronic files or apps. The school says the aim is to develop “AI-resilient” teaching that requires students to think critically and independently rather than rely on automated assistance.
Chicago is not excluding AI from legal education entirely. Its first-year legal research and writing programme is designed to begin with independent writing before introducing AI for research, revision, draft iteration and preparation for oral argument. Upper-level courses have greater flexibility, and instructors must state their AI rules in their syllabuses.
Berkeley Law has taken a restrictive default approach to assessed work. Its policy, effective from summer 2026, prohibits students from using AI to conceptualise, outline, draft, revise, edit or translate work submitted for credit. AI is also prohibited in examinations, and students may not upload course materials to generative AI systems. Faculty can adopt different rules where they consider AI use appropriate for the course.
Columbia Law School has adopted a more permissive framework. Students may use AI for research, to test arguments, seek criticism, explore alternative formulations and improve clarity, but they may not rely on it to perform the original legal reasoning or analysis an assignment is intended to assess. Work submitted for credit must remain student-composed, while AI use during examinations is prohibited by default unless an instructor expressly permits it. Columbia is also introducing AI training during student orientation and embedding AI modules in selected first-year courses.
The University of Georgia School of Law is also combining restrictions with technology training. Its 2026-27 handbook prohibits classroom use of electronic devices unless authorised by an instructor or required as an accommodation. Students matriculating from autumn 2026 must also complete a course satisfying a new technological competence requirement before graduation.
The scale of the policy shift is illustrated by an archive created by Suffolk University Law School dean Andrew Perlman. The database, updated on 20 September, contains public-source material from 180 U.S. law schools and currently lists 38 schools under its mandatory-curriculum category. The archive also tracks policies governing exams, submitted work, classroom use and faculty support.
The changes have UKJT AI Liability Statement — Lawyer Monthly beyond the classroom. What is the opinion on AI in law says lawyers using generative AI must consider existing duties including competence, confidentiality, supervision, client communication and candour to tribunals. The guidance reflects the increasing expectation that UKJT AI Liability Statement — Lawyer Monthly forms part of modern legal practice.
Law schools are therefore confronting two competing demands: ensuring students develop legal judgment without outsourcing foundational reasoning, while also preparing graduates to use AI responsibly once they enter practice. The emerging policies suggest that the central question is increasingly not whether future lawyers will use AI, but when, for which tasks and under what safeguards.