Using AI in Law School Without Violating the Honor Code
AI in law school isn't going away, and pretending you don't use it isn't a strategy — it's a landmine. The honest question is which uses are within your school's honor code and which aren't. The answer is more permissive than most students assume, and less permissive than the loudest voices on Reddit claim.
What honor codes actually prohibit
Every JD honor code we've reviewed prohibits some version of three things: (1) submitting work that isn't substantially your own, (2) using unauthorized aid on graded assignments, and (3) misrepresenting the source of your work. AI is a new kind of aid, but the rules that govern it are old.
The specifics matter. Some schools (Yale, Michigan, Georgetown) have published AI-specific guidance. Others fold AI into their existing 'unauthorized aid' rules. Read yours before you decide what's safe.
Four uses that are almost always safe
- Explaining a case or doctrine to you. Asking an AI to explain the rule against perpetuities in plain English is the same as asking a study group or reading a hornbook. You're being taught, not writing.
- Quizzing you on rules and elements. Flashcard drills, spaced repetition, and Socratic-style questioning are all study aids — they don't produce submitted work.
- Summarizing your own notes back to you. If the AI is only working from material you produced, it's a study tool, not a ghostwriter.
- Formatting and citation help. Bluebook fixes, converting rough notes into an outline structure, and grammar checks are indistinguishable from Word's spell-checker.
Three uses that get people in trouble
- Drafting graded work. Any essay, memo, brief, or exam answer that will be turned in for a grade must be substantially your own writing. AI as a co-author of graded work is where honor-code cases live.
- Fabricating citations. AI models hallucinate case names and citations. Submitting AI-generated case citations without independently verifying them in Westlaw or Lexis has already produced sanctioned lawyers — you're not immune.
- Getting the answer to a take-home exam. If your professor bans outside aid, AI is outside aid. Period.
The disclosure principle
When in doubt, disclose. A short 'I used [tool] to [specific task]' note at the top of an assignment converts an honor-code question into a professor-preference question — and professors are generally reasonable when you're upfront. Silent AI use on graded work is where trouble compounds.
The verification principle
AI is confidently wrong about legal citations at a rate that would end a lawyer's career. Every case name, every citation, every rule statement an AI produces has to be verified in a real legal database before it appears in anything you submit. This isn't optional.
Where Scriba fits in
Scriba is built around these two principles. The AI assistant refuses to draft graded work outright. Every AI answer is grounded in passages you highlighted, and unverified case citations are flagged in red until you check them. Every export ships with an AI-disclosure appendix listing which passages the assistant touched — so if your professor asks, you have the receipts.
Frequently asked
Almost always yes. Outlining is study, not submission. Just don't paste an AI-generated outline into a graded assignment and call it your own.
Then don't use it for that class's graded work. You can still use AI as a study aid for your own comprehension — that's not banned, that's how you learn.
A one-sentence note at the top of the assignment is standard: 'I used [tool] to [task, e.g. quiz me on the elements of negligence].' Scriba's export includes a disclosure appendix you can attach.
Scriba is a notebook-style desktop workspace for law students. 7 days free, cancel any time.
Start free trial