Berkeley Law’s strict default rules on artificial intelligence have drawn criticism from a student AI group, bringing a question into focus for legal education: how should schools test independent legal reasoning while preparing graduates to use AI in practice?
The policy, effective in summer 2026, prohibits students from using AI to conceptualize, outline, draft, revise, translate or edit work submitted for credit. It also bars AI use in exams and the upload of course materials to generative AI systems. Instructors can set different rules with written notice, and students must disclose authorized use. The default rule allows AI in paper research only to identify sources, which students must check themselves.
This is not a campus-wide AI ban. Berkeley Law says instructors may allow AI when it suits a course’s learning goals, and the school offers AI and technology law courses. Nor did Berkeley announce the classroom laptop and phone bans described elsewhere in a September 21 Reuters report; those device restrictions are policies at other law schools.
What did the students object to?
Reuters reported that the student-run group AI at Berkeley Law challenged the default restrictions as difficult to reconcile with the school’s leadership in law and technology. The group drew a distinction between handing legal reasoning over to a machine and learning to use AI within a legal workflow. Professor Colleen Chien, a member of Berkeley Law’s AI Leadership Committee, told Reuters that faculty can depart from the default where their teaching goals call for it.
The disagreement is about which skills students should demonstrate unaided and when supervised AI practice belongs in the curriculum. Berkeley Law’s policy explanation says the school wants exams and papers to measure students’ own analysis and writing. It also says students need to learn to use AI effectively. A grammar correction or a proposed paper outline may feel routine in a workplace, but both are prohibited under Berkeley’s default rule for credited work.
What could law schools do next?
A workable approach would state the AI rules for each assignment, assess foundational reasoning without AI where appropriate, and create separate exercises in which students must verify an AI-generated result against primary legal sources. That is an editorial proposal, not a description of a new Berkeley requirement.
For Philippine law schools and legal employers, the US dispute offers a useful question rather than a rule to import: can a trainee identify a fabricated case, protect confidential material and explain a legal conclusion independently before relying on an AI-assisted workflow? Local institutions would need to set their own assessment and tool-use policies.
Sources: Berkeley Law AI policy; Berkeley Law policy Q&A; Reuters, September 21, 2026.
Illustrative photo: freddie marriage / Unsplash.