One of the world's top law schools draws a hard line against AI in legal education

The policy's carve-out for research use is doing a lot of quiet work here: Berkeley isn't anti-AI, it's making a deliberate bet that tool access without foundational skill is a liability, not an asset. That distinction matters because it reframes the debate from 'AI in education' to 'sequencing of AI exposure.'

This story sits in a different lane from recent coverage like The Verge's piece on Google's anything-to-anything model, which focused on capability expansion at the consumer and creative layer. Berkeley's move is less about what AI can do and more about what institutions do when capability outpaces pedagogy. The relevant context is the broader pattern of knowledge-work institutions trying to define where human skill ends and tool augmentation begins, a tension that has no clean resolution in the current moment. Berkeley is essentially running a controlled experiment on whether restricting access produces better-reasoned graduates, and the results will matter to any firm that hires from its pipeline.

Watch whether peer schools (Yale Law, Harvard Law, NYU) issue explicit AI policies within the next two academic cycles. If two or more adopt similar research-only carve-outs, Berkeley's framework becomes a de facto standard rather than an outlier position.