The Berkeley Law Red-Light AI Policy

For months now, I've been looking for examples of program-level responses to generative AI at colleges and universities. Almost all the work I've seen adapting to the challenges and opportunities that AI poses to teaching in higher ed has been at the level of the individual course. That's a great place to practice AI-aware teaching, of course, but at some point, our students will need more coherent approaches to AI across the courses they take.

Last week the University of California Berkeley School of Law announced a school-wide AI policy that, if nothing else, provides that sense of coherence for students. The school's new AI policy reads, in part:

Instructors may set alternative policies for individual courses and assignments, but the default policy for Berkeley Law courses is now a fairly strict "red light" policy.

There's definitely value in students knowing what to expect in terms of AI policies across their courses. Without a schoolwide policy like this, it's likely that a student would take courses with wildly different AI policies. Within a semester that means potential confusion for students about which uses of AI are allowed for different courses, and across semesters that means students might be taught how to use AI effectively for certain tasks in one course only to be told by a subsequent instructor not to use AI for those tasks. Having a blanket policy, whether it's a red-light or a green-light policy, avoids these challenges. It also likely makes enforcement of academic integrity policies simpler and more manageable.

I appreciate that the new Berkeley Law policy includes a clear rationale for the new restrictions. "Future lawyers may need to use AI fluently," the policy states, "but the current state of the technology requires that AI use be coupled with the cognitive skills necessary to strategically deploy the technology, to critically assess its work product, and to uphold ethical obligations to clients and to the legal system." The policy argues that developing those skills and associate professional judgment won't be served by student AI use, and thus student AI use is prohibited.

I'm not a lawyer or a legal educator, so I can't say if the argument here is the right one, but it certainly seems reasonable, given that practicing lawyers continue to submit legal documents with fake citations generated by AI. There's an online database of these incidents maintained by Damien Charlotin that has identified 1,497 such cases, including 41 just this month. Clearly there's a need for at least some lawyers to bring a little more expertise to their use of AI.

The Berkeley policy reminds me of a course-level policy I learned about recently from one of my University of Virginia colleagues. Daniel Radthorne is a research librarian at the UVA School of Law and served as one of UVA's 2025-2026 Faculty AI Guides. He was one of several Faculty AI Guides whose AI syllabus statements I was able to share in a new UVA Teaching Hub collection. His course policy is more of a yellow-light policy, but he offers a rationale similar to the one from Berkeley:

This approach makes a lot of sense to me. Leave the AI out of the work where it might derail the kind of conceptual understanding and knowledge organization needed to do legal research, but also teach the AI-powered tools that are becoming widely used in that legal research.

It's a little odd to me that the Berkeley policy is aimed at developing the cognitive skills that lawyers need ("thinking remains the sine qua non of good lawyering," as the policy states) while singling out the use of AI to identify sources as the only broadly acceptable use of AI by students. Finding legal sources is where most of those 1,497 lawyers in Damien Charlotin's database tripped up, and the need to think critically about legal sources is why Daniel Radthorne limits his students' AI use. I think there's a way to square this circle (there's plenty of opportunity for critical thinking about sources once they've been identified), but I wonder how Berkeley students might respond to the carve-out for AI use in legal research. Will it lead them to believe that legal research doesn't require the same kind of thinking as other legal practices?

Another concern I have is how a red-light policy like Berkeley's will be implemented at a time when just about every piece of software in use is adding some kind of AI feature. I really like that the Berkeley policy details specific uses that are banned--asking an AI tool to brainstorm a paper topic, asking an AI tool to identify repetitive passages in a paper that should be cut, asking AI to generate an exam outline for use on an exam, and so on. These are tool-agnostic examples that clarify the boundaries of appropriate and inappropriate AI use for students, and this is a case where clear is kind. But with AI tools intruding all the time (not unlike Microsoft Clippy's "It looks like you're writing a letter!"), I can imagine students worried about policy violations turning to pencil and paper for all their course work, which comes with its own set of problems.

A red-light policy like this might also lead some students to go underground with their use of AI tools. It's still very hard to successfully detect and prove unauthorized AI use by students, so at least some of these students will likely "get away" with prohibited AI use. That poses fairness problems if those students get ahead somehow of other policy-following students. Also, as I wrote earlier this month, there's research indicating that when students aren't guided toward productive uses of AI in their studies, they tend to use AI in ways that undercut learning. The Berkeley policy makes it less likely that students will receive that kind of guidance from faculty.

There are also some prohibited use cases that seem perhaps overly restrictive. For instance, on the banned list is "asking AI to translate a paper originally written in another language into English," which is aimed, it seems, at preventing students from writing a paper in their preferred language then having AI translate that paper into English. Developing "fluency with legal English," as the policy states, seems like a reasonable goal, but I wonder what other kinds of AI-powered translating potentially useful for English language learners would be prohibited. What about the use of something like Goblin Tools to assist with time and project management? I think that's permitted by the Berkeley policy, but using AI tools like Speechify to turn course readings into spoken word--a very useful accessibility move for many students--would be prohibited.

I'm glad to have an example of a school-wide AI policy like this to analyze, and I welcome your thoughts on the Berkeley Law policy. And if you know of other program-level responses to generative AI, please share! I'm particularly interested in how programs are revising learning outcomes because of the ways that AI is changing work and information literacy.

Teaching Takes with Bryan Doppel

I had a lot of fun talking with Bryan Doppel, education development specialist at Appalachian State University, while recording the latest episode the Teaching Takes podcast. Bryan has an engaging format for the show where he and I commented on "takes" on the episode topic (AI and teaching, naturally) submitted by listeners. We talked about the ways that clear course AI policies can help students understand how to use AI to support (and not shortcut) their learning, the value of "critique the chatbot" assignments for developing critical AI literacy, and the need for curriculum-level (not just course-level) responses to generative AI.

As Bryan said on LinkedIn, "this is an alarmingly and aggressively reasonable conversation about student and instructor AI use." You can listen to our conversation and other episodes of the Teaching Takes podcast here.

The Norton Guide to AI-Aware Teaching

My new book, The Norton Guide to AI-Aware Teaching, co-authored with Annette Vee and Marc Watkins, is now available to pre-order! The ebook is expected to be available July 1st, and print copies are expected to start shipping on September 24th. Here's how you can get a copy:

1. Our publisher Norton is pleased to offer the guide as a free ebook for all instructors currently using a Norton textbook. If that's you, you'll receive access from the Norton team when the ebook is available July 1st and can contact your local Norton representative with any questions.
2. If you would like to pre-order the ebook so that you have it July 1st, you can now do so through Amazon and Barnes & Noble and perhaps other retailers.
3. If you would like to pre-order the paperback version of the book, you can now do so through Norton, Amazon, Barnes & Noble, and likely other retailers. If you go through Norton, be sure to use the code AIFREESHIP at check out to get free shipping!
4. If you would like to order multiple copies for a campus reading group or some other faculty development effort, Norton has an option for you: On orders of 10 or more print copies, we offer 50% off the list price and free domestic shipping. (Such orders must be on a nonreturnable basis.) To take advantage of this offer, contact Peter Wentz at pwentz@wwnorton.com with subject line “Norton Guide to AI-Aware Teaching.”