As law schools finalize their AI policies for the current academic year, two prominent institutions illustrate just how differently legal education is approaching this question. Columbia Law School's newly adopted policy doesn't impose a blanket ban, instead requiring that students retain genuine intellectual responsibility for credit-bearing work: AI can serve as a learning aid or a tool for testing ideas, but it cannot substitute for a student's own legal reasoning or serve as an undisclosed ghostwriter, and its use during examinations is generally prohibited outright. At the University of California, Berkeley School of Law, the current policy takes a considerably more restrictive stance, prohibiting students from using AI to conceptualize, outline, draft, revise, translate, or edit any work submitted for academic credit.

This is a genuinely significant divergence between two highly regarded law schools, and it reflects a real, unresolved tension specific to legal education: legal writing itself, structured heavily around frameworks like IRAC that reward precise, rule-governed analysis, is in some ways a genre AI tools handle with real facility, which raises the stakes on academic integrity concerns in a way that may feel sharper in legal education than in some other academic fields where AI's current limitations are more readily apparent in the writing it produces.

At the same time, legal practice itself is rapidly incorporating AI tools professionally, which creates a genuine pedagogical tension law schools are navigating differently: preparing students for a profession where AI-assisted legal research and drafting tools are increasingly standard, while ensuring students develop the underlying legal reasoning skill that using those tools well ultimately depends on.

What this divergence suggests for legal writing instruction

Neither approach is obviously correct, and the divergence itself is instructive: it suggests legal education hasn't yet settled on a shared consensus about how to balance AI's genuine professional relevance against the specific integrity risks it raises for a discipline built so heavily around structured, rule-governed written analysis. Law schools and individual legal writing instructors navigating this uncertainty have real reason to look at both models closely rather than assuming one institution's approach represents an emerging consensus the rest of legal education will inevitably converge toward.

• Review your own institution's current AI policy language carefully, since the range this year spans from permissive to quite restrictive
• Consider how IRAC-structured legal writing's rule-governed nature specifically shapes AI integrity concerns compared to other writing genres
• Weigh the tension between preparing students for AI-assisted legal practice and ensuring they develop underlying reasoning skill independently
• Watch how these differing policy approaches perform over the coming year as an early comparative case study
• Communicate your specific institution's policy clearly to students, given how much variation currently exists across law schools

What grading needs to look like under either model

Regardless of which policy model a given law school adopts for student AI use, grading legal writing itself, IRAC-structured exam answers and legal memos, remains a distinct question from student-facing policy, and one where rubric-based, component-by-component evaluation continues to offer real consistency benefits for instructors and graders, independent of how permissive or restrictive the underlying student use policy happens to be. A law professor grading exam answers against a shared, structured rubric is addressing a genuinely different question than the one these student-use policies are trying to answer.

This distinction, between policy governing student use of AI in producing their work and the separate question of how faculty and graders evaluate that work once submitted, is worth keeping clear as legal education continues working through what remains a genuinely unsettled policy landscape this year.

Where this leaves legal writing programs heading into the year

With no clear national consensus emerging yet among leading law schools, legal writing programs have real latitude, and real responsibility, to think through their own specific policy carefully rather than assuming any one peer institution's approach represents settled best practice. The genuine divergence between institutions as prominent as these two suggests this is a question legal education as a whole is still actively working through, not one where the answer has already been found.