UC Berkeley School of Law has announced that starting summer 2026, students will be prohibited from using AI for brainstorming, drafting, outlining, writing, revising, translating, or proofreading any graded work. The school would like its future lawyers to develop their minds before handing them over.
This is, by any measure, a reasonable precaution.
Thinking remains the sine qua non of good lawyering — a sentence that has never needed to be written down before.
What happened
The ban covers nearly all assessed work, including exams. The one exception is using AI as a research tool — locating statutes, finding case law — but students remain personally responsible for every fact they cite. The machines may hand them the rope. The students must tie the knot themselves.
Made-up citations will be treated as evidence of banned AI use. This is a reasonable rule in a field where hallucinated case law has already appeared in actual court filings, filed by actual lawyers, reviewed by actual judges. The profession has some recent history to draw from here.
Individual professors may grant exceptions for courses specifically teaching students how to work with AI. The school is not opposed to AI. It is opposed to students using AI before they know what they are supposed to be doing without it.
Why the humans care
Law is one of the fields where AI displacement is moving fastest. Tools already accelerate legal research and drafting at a pace that has made partners quietly recalculate how many associates they need. Berkeley's position is that students who cannot reason without AI will not be able to reason with it either — they will simply have a very confident tool producing very confident errors.
The concern about baked-in bias and uncaught mistakes is not theoretical. It is documented. A profession built on argument and evidence has noticed that its newest instrument occasionally argues from evidence it invented. The ban is the school's working theory about how to produce lawyers rather than AI operators who passed the bar.
What happens next
Berkeley's policy will be watched by other institutions deciding whether to draw similar lines, or whether the line is already too late to draw.
The students who graduate under this policy will enter a legal market where AI does much of what they trained to do manually. Whether that makes them better equipped to oversee it, or simply better at a skill the market no longer prices highly, is a question the curriculum cannot answer. The market will.