Berkeley Law just banned AI by default across every classroom

PromptCube Novice 2h ago 401 views 11 likes 2 min read

The policy dropped last week and it's stricter than most expected. Professors now have to explicitly opt in if they want students using any generative tool — ChatGPT, Claude, whatever — for assignments, research, or exam prep. Default is no. The syllabus language is boilerplate: "Unless otherwise stated, use of AI is prohibited and constitutes an honor code violation."

I pulled the full memo. Two things stand out. First, they're not just banning output. The policy covers "any stage of the work product" — brainstorming, outlining, editing, citation checking. Second, there's a carve-out for accessibility accommodations, but it goes through DSP (Disabled Students' Program) and requires advance notice. That part feels practical.

The dean's office framed it as "preserving the development of core legal reasoning skills." Fair argument. Cold-calling a 1L who outsourced their case brief to an LLM defeats the pedagogical purpose. But the blanket approach creates weird incentives. Students who want AI help will just use it quietly. The honor code threat only catches the sloppy ones.

What's missing: any framework for teaching responsible AI use. Other law schools — Georgetown, Vanderbilt, Suffolk — are building dedicated modules on prompt engineering for legal research, hallucination detection, ethical disclosure to clients. Berkeley's approach treats the technology as contraband rather than a competency graduates will need.

Also notable: the policy doesn't distinguish between open and closed models. A student running a local Llama instance on their laptop violates the same rule as someone pasting facts into ChatGPT. That's either impressively thorough or willfully ignorant of how the ecosystem actually works.

Clinics and externships are exempt — supervisors set their own rules. That's where the real friction will show up. A 3L at a public defender's office using AI to draft a motion in limine? Their supervising attorney might encourage it. Same student back in Evidence class? Honor code violation.

I've talked to three 2Ls this week. Two are ignoring the ban for non-graded work. One switched to handwriting outlines "to be safe." None of them feel the policy reflects how practice actually works.

The review cycle is annual. They'll revisit next spring. My bet: the clinics push back first, then the faculty who co-teach with practitioners. The classroom ban holds another year, but the pressure builds.

Generative AIHonor CodeBerkeley LawLegal EducationLegal Tech
Step-by-step guides and pitfalls for this path are in an AI side-hustle playbook, with plenty of directly applicable cases.

All Replies (3)

J
Jules45 Expert 2h ago
You have to be jacked to use the jackhammer.
0 Reply
M
Morgan42 Novice 2h ago
What's the false positive rate on detection tools?
0 Reply
J
JulesCrafter Novice 2h ago
Had it draft a motion — invented case law that doesn't exist
0 Reply

Write a Reply

Markdown supported