Is ChatGPT Good for Law School? An Honest Answer
Yes for some things, no for others, and the difference is not obvious from using it. ChatGPT is excellent at the task law students reach for it least and unreliable at the task they reach for it most.
Here is the split, why it falls where it does, and the policy question worth settling before you rely on any of it.
Where It Is Genuinely Good
Explaining a concept you did not follow in class. This is the strongest case and it is not close. Ask why personal jurisdiction has the shape it does, or what separates a condition from a covenant, and you get a clear explanation you can ask follow-up questions about at your own pace. No citation is required, nothing is submitted, and the failure mode is a slightly imprecise analogy rather than a fabricated case.
Restructuring your own writing. Paste your outline section and ask it to tighten the hierarchy. Paste a paragraph from a memo and ask where the reasoning skips a step. Working on text you wrote is safe, because the substance is already yours.
Generating practice hypotheticals. Issue spotting improves with volume, and the bottleneck has always been finding enough problems. A chatbot removes that bottleneck. Ask for ten negligence hypos with buried issues and you have a week of reps.
Playing the other side. "Write the strongest argument for the defendant here" is a genuinely useful prompt. It surfaces the counterargument you would have missed, which is most of what exam grading rewards.
Where It Quietly Hurts
Anything requiring a citation. ChatGPT invents cases. Not occasionally, predictably: it is trained on the open web rather than a legal database, and it is built to produce plausible text. Asked for authority, it generates a real-looking case name, a correctly formatted citation, and a holding that fits your question. Often the case does not exist.
This is not theoretical. In 2023 two New York attorneys were sanctioned in federal court after filing a brief built on cases the model invented, and courts in multiple jurisdictions have handled similar incidents since. For a student the parallel is a fabricated cite in a graded memo, and most schools treat that as academic misconduct whether or not you knew.
Briefing cases you have not read. It does not have your casebook. Casebook opinions are heavily edited, and the model may reconstruct from the full opinion, a different version, or an approximation. The brief reads fine and then a cold call asks about the procedural posture.
Test it yourself in about ten seconds
Ask ChatGPT for the holding in a case you know cold, then search the same case in Case Cub. One answer is generated from training data. The other is built from the actual opinion, with the citation, the posture, the key terms, and the precedent the court relied on. No account needed to see the difference.
Writing your outline. It will produce a competent generic outline organized like a treatise. Your exam is organized like your professor's syllabus, weighted toward whatever they spent three weeks on. An outline that is not yours also skips the part where making it is how you learn the material.
Grading your practice answers. It rewards answers that look complete over answers that spot the hidden issue, which is the opposite of how exams are actually graded.
The Policy Question
Check your school's academic integrity policy and each syllabus before you settle into a workflow. Policies vary widely right now, and several patterns are common: AI permitted for study but not for graded work, permitted with disclosure, banned for anything submitted, or left entirely to individual professors. Some professors ban it in one class and encourage it in another.
Two things are true almost everywhere. First, submitting AI-generated text as your own is misconduct. Second, a fabricated citation is a problem regardless of the tool that produced it, because the obligation to verify is yours.
If the policy is unclear, ask. It is a much better conversation to have in September than in front of a committee in April.
What to Use Instead, By Task
- Reading and briefing: something grounded in real opinions rather than generating them. Our case brief library is built from actual cases, and the case brief template is free if you would rather write your own.
- Verifying any citation: Google Scholar, CourtListener, or your school's Westlaw and Lexis access, which is free to you now and expensive later.
- Outlining: your own notes, with AI used only to compress language you wrote.
- Exam prep: generated hypos for volume, a rubric you built from the syllabus for grading.
A fuller task-by-task breakdown is in the best AI for law school, and the tool comparison is in best AI tools for law students.
The One Sentence Version
Use ChatGPT to understand things, never to source things, and never for anything you submit without checking every citation in a real database.
Related
- Best AI Tools for Law Students
- Case Brief Generator, search 20,000+ briefs written from real opinions
- How to Study for Law School Exams