How this site uses AI
Core Legal Rules — public transparency statement. Last updated 2026-06-21.
Most products that use AI treat it as a black box. You know it's there; you're asked to trust it; you're shown nothing. We do the opposite. This page, and the more detailed methodology behind it, exist so you can see exactly how this site is built and decide for yourself how much to rely on it.
The short version
The legal rules on this site were drafted with AI, organized with AI, and are being verified with AI. A person is accountable for the integrity of that process — for telling you the truth about how it works and where it can fail — but no person certifies that every rule is a perfect statement of the law. We don't claim that, because it wouldn't be true. What we claim is narrower and, we think, more useful: that our method is disclosed, our sources are linked, our checks are real, and our error rate is something we measure rather than something we assert.
We checked whether anyone gets this right
It may surprise you how often errors turn up even in the most respected scientific journals — Nature, Science, and the rest; Stuart Ritchie's book Science Fictions documents it in unsettling detail. Because we care so much about whether you can rely on what we publish, we didn't simply assume the law was different — we looked into it. The finding is humbling rather than comforting: the law appears more disciplined about citation-checking than the sciences, but — tellingly — it has rarely measured its own error rate at all, so the absence of a number is not the absence of errors; and even careful legal scholarship struggles most with making corrections visible and durable. No serious knowledge system is error-free.
That didn't discourage us — it told us what to build: a system designed to surface, source, check, and openly correct errors, rather than one that asks you to assume there are none. How we do that is below, and the fully-sourced version is in our methodology page.
What that means for you
This is a public beta. The rules are a study and reference aid, not legal advice, and not a substitute for the controlling primary authority. Where a rule rests on a statute, federal rule, or constitutional provision, we link you to the official text so you can read it yourself. Where a rule reflects common-law doctrine, we say so and frame it as the majority or exam-tested rule, which can differ from the law of any particular state.
The Multistate Bar Examination tests a specific, conventionalized version of the law. Our rules are written to that convention. That is a deliberate simplification, and it is one more reason to verify against primary authority before relying on any statement here.
When we show you a court case, a computer found it — an AI did not
You may have read about lawyers being sanctioned for citing cases that an AI made up. That failure — an AI "hallucinating" a plausible citation to an opinion that does not exist — is the most notorious risk of using AI in law, so it is worth being plain about how we avoid it.
There is a difference between automation and artificial intelligence, and we use them for different jobs. Automation is a computer following fixed instructions to retrieve real records that already exist — closer to looking a word up in a dictionary than to writing an essay. Generative AI writes new text and can be confidently wrong. We never let the writing kind of AI invent authority.
So when a rule shows "Selected representative cases," those cases were not chosen or written by an AI. A program searched CourtListener — a free, public database of millions of actual court opinions — for decisions that state the rule in their own words, and recorded the real results. Every case links straight to the opinion so you can open it and read it yourself. They are real because they were retrieved from a database of real decisions, not generated.
What automated retrieval guarantees is that the opinion exists and uses the rule's language; it does not by itself certify that the case is the best or most current authority in your state, and these selections are still moving through our review. As always, read the linked opinion and verify against primary authority before you rely on it.
If you see something wrong, tell us
We are not crowdsourcing these rules — this isn't a wiki, and users don't edit the content. But motivated readers catch real errors, and we want to know. Every rule has a way to flag a problem or ask a question. Every report we receive is reviewed, and disputes feed directly back into the verification process that checks the rest of the corpus.
The short answer
Honestly: verify before you rely. These rules were generated and are being verified by AI, and they're in public beta — most have not yet completed our full verification process. They're a strong, fast way to recognize and recall the rules the bar exam tests, and wherever we can, we link you straight to the official source so you never have to take our word for it. Treat the site as a sharp study aid that shows its work, not as an authority of last resort. If you find a mistake, there's a button on every rule to tell us — and we read every one.