UP’s AI ban gets the training question backwards
A first-year ban on generative AI may look cautious, but lawyers will not be entering a profession that lets them avoid the tool, only one that punishes them for trusting it blindly.
An AI-assisted editorial, reviewed by a human before publishing. It reasons over our own tracker data (and cited context) — a point of view, not legal advice.
The conventional wisdom is simple: keep generative AI out of first-year law school work and you will protect students from shortcuts, sloppy thinking, and fake citations. That sounds clean. It is also the wrong instinct if it becomes the whole policy.
The University of the Philippines College of Law has adopted a new rule for first-year JD courses starting in AY 2026-2027. Core subjects are to be taught offline where possible, generative AI is banned in writing assignments, and only limited faculty-supervised use of AI-based research tools is allowed, with disclosure. That is the peg. It is also a signal that legal education is starting to pick a side on how early students should meet the machine.
AI can draft fast and lie faster.
A year ago, the answer looked easier
A year ago, the easy answer was that law schools could still treat AI as a future problem. Today that answer is gone. One recent archive from Suffolk Law Dean Andrew Perlman says it catalogs public AI policies, teaching strategies, and curricular programs from 128 of the 196 U.S. law schools. That alone tells you the argument has already moved from whether schools should respond to what they should make students do.
The split is already visible. A recent opinion in Inside Higher Ed says the University of Chicago Law School is taking a gradual approach, keeping foundational first-year work without AI and layering supervised use later. By contrast, the University of California, Berkeley School of Law drew attention for a highly restrictive policy that broadly prohibited generative AI in graded work. UP has now landed on the restrictive side for first-year core work. The question is whether that restraint teaches judgment, or just delays the moment of reckoning.
What changed is the profession, not the students
The market has stopped pretending this is fringe. Law.com reports that 132 people now hold AI ownership roles across law firms, with many firms naming chief AI officers or assigning AI responsibility to innovation or knowledge leaders. Latham & Watkins says it has launched an AI Academy for its lawyers. Thomson Reuters and Laurel have announced a partnership aimed at measuring AI use against timekeeping, profitability, and return on investment. Microsoft says it will implement Harvey across its corporate, external, and legal affairs operations. LexisNexis is rolling out new AI features across drafting, research, and workflows.
If you are a working lawyer, that matters more than any school policy. The tool is already in the office, in the vendor stack, and in the management conversation. A classroom ban does not stop a junior associate from entering a firm where leadership expects fluency, or a client from expecting faster turnaround because the firm says it has invested in the tool. It just means the student may arrive with a cleaner conscience and a weaker habit of checking output before trusting it.
Blank prohibition teaches the wrong reflex
The strongest argument against a blanket ban is not that students should be handed every shiny tool on day one. It is that lawyering now requires controlled use, verification, and citation discipline. Those are habits, and habits are built early. A rule that says generative AI is simply off limits in core writing assignments teaches avoidance. That is not competence.
The record of what happens when lawyers skip the checking step is already ugly. There are verified sanctions matters where fake authorities made it into court filings, including a U.S. 2023 sanctions matter in which two lawyers and their firm were fined $5,000 each, and another 2023 Southern District of New York case where a lawyer was challenged over seven apparently fictitious cases and the court imposed a $5,000 fine on two lawyers and their firm. In Harber v Commissioners for His Majesty’s Revenue and Customs, a litigant in person filed an appeal submission citing nine non-genuine authorities hallucinated by generative AI. The lesson is not that AI is forbidden. The lesson is that unverified AI is a professional hazard.
Teach the use, or the market will
This is where UP’s limited exception matters. The policy still allows faculty-supervised use of AI-based research tools with disclosure requirements. That part is the right instinct. It recognizes that students need to learn where the tool can help and where it can mislead. The mistake is drawing the line so sharply around writing that students may never practice the skill that now matters most: deciding what to trust, what to cite, and what to verify before they put their name on it.
Legal education should not be trying to produce graduates who have never touched AI. It should be producing lawyers who know that AI can draft fast and lie faster. The profession is already building formal leadership around it, from chief AI officer roles to firmwide academies, because the clients, the vendors, and the internal economics are pushing in that direction. If schools answer that pressure with a total ban in the places where judgment is most tested, they will graduate students who understand the rulebook and arrive underprepared for the practice. That is not caution. That is deferral, and the bill comes due the first time a junior lawyer signs off on something they did not really check.