UC Berkeley law bans AI use

UC Berkeley Law has adopted a new default rule that bars students from using artificial intelligence for most graded academic work, tightening a campus policy that had previously allowed narrower uses such as grammar correction and research. The law school’s policy page says the rule takes effect in summer 2026 and prohibits AI for “conceptualizing, outlining, drafting, revising, translating, or editing” any work submitted for credit. It also bans AI “for any purpose in any exam situation” and forbids students from uploading course materials into generative AI systems. Berkeley Law says instructors may depart from the default only in courses intentionally designed to teach AI fluency or where a professor sets a different rule. ### How broad is the Berkeley rule? Berkeley Law’s published rule reaches well beyond banning AI-written final drafts. The policy says students may not use AI to brainstorm a paper topic, propose an organizational structure, summarize a legal rule for use in a paper, identify repetitive passages to cut, correct grammar, translate work into English, or generate exam outlines later used in an exam. (law.berkeley.edu) The school’s policy page says AI may still be used for paper research, but only “for the limited purpose of identifying sources, such as cases, statutes, or secondary sources.” The same page says students remain responsible for the accuracy of that research and for all other aspects of submitted work. (law.berkeley.edu) ### What reason did the law school give? UC Berkeley Law framed the rule as a training issue as much as an integrity issue. The policy says future lawyers may need to use AI fluently, but that current tools must be paired with the cognitive skills needed to deploy them strategically, assess their output critically and meet ethical duties to clients and the legal system. (law.berkeley.edu) The policy page says the school wants students to develop their own ability to “conceptualize, outline, draft, revise, and edit” legal work. It adds that the rule is meant both to support “the best legal education possible” and to promote “fairness and administrability.” ### How does this differ from Berkeley’s earlier approach? (law.berkeley.edu) A Reuters report carried by the Association of American Law Schools said Berkeley Law’s April 14, 2023 policy allowed students to use AI for research or grammar correction, while barring its use on exams and in composing submitted assignments. That earlier rule was presented as one of the first formal law-school policies on student use of generative AI. (law.berkeley.edu) The summer 2026 policy narrows that permission. Grammar correction is now listed among prohibited uses for work submitted for credit, and the new rule spells out a longer list of barred activities tied to the full writing process. ### Are there any exceptions? Berkeley Law’s policy leaves room for faculty discretion, but only through explicit course-level exceptions. (aals.org) The rule says instructors may deviate from the default in courses designed to teach AI fluency or in other courses where a different rule is “pedagogically appropriate.” (law.berkeley.edu) That means the school’s baseline is prohibition for graded work unless a professor affirmatively says otherwise. The policy is posted on Berkeley Law’s academic rules page and in a PDF labeled effective summer 2026. ### What should students watch for next? Summer 2026 is the operative date in Berkeley Law’s posted materials. (law.berkeley.edu) Students will need to check individual course rules for any professor-approved exceptions, while the default policy on the law school’s registrar page governs graded work, exams and the handling of course materials in generative AI systems.