For Australian law students
CaseSharp and ChatGPT
You can already ask a general assistant about Australian law, and plenty of law students do. That is a reasonable thing to do. The difference is not that one is smart and the other is not. It is where the answer comes from and whether you can check it without doing the work twice.
Published 4 August 2026. Written by the CaseSharp team in Sydney.
Short answer
Use a general assistant to understand something. Use CaseSharp when the answer has to carry a citation you will be marked on. A general model answers from what it learned. CaseSharp answers from a corpus of audited Australian decisions searched at the moment you ask, links every claim to the paragraph it came from, and shows how the case has been treated since.
Side by side
| The job | CaseSharp | ChatGPT |
|---|---|---|
| Where the answer comes from | A corpus of 73,000 audited Australian decisions, searched at the moment you ask | A general model, answering from what it learned in training plus anything it retrieves in the session |
| Checking a citation | Every citation is a link to the paragraph it came from. Click it and you are reading the judgment | You copy the case name into AustLII or Jade and check it yourself, one at a time |
| When there is no authority | Sharp says it cannot find support and shows you what it did find | It will usually still produce an answer |
| Citation graph | Every authority carries a citation graph: which cases cite it and which it cites, ranked by influence, each one linked so you can open it | No Australian citation graph by default, so it cannot show you which later cases have cited an authority |
| AGLC4 | Full form, short form and pinpoints, flagged for manual review when a reference cannot be parsed | Formatting varies between answers and pinpoint rules are not reliable |
| Australian coverage | Australian decisions and legislation only, which is the point | General purpose across every jurisdiction. Australian material sits alongside everything else rather than being the whole of it |
| Cost | A free plan that does not expire and does not need a card, or A$49 a month, A$470 billed annually | There is a free tier and there are paid tiers. Check ChatGPT's current Australian pricing on their site |
Where ChatGPT is still better
Explaining a concept from scratch, drafting anything that is not legal writing, working through non-legal subjects, and thinking out loud at 1am. It is a better generalist than CaseSharp will ever be. What it does not do is hand you the paragraph.
What a pinpoint citation actually is
A pinpoint is the number that comes after the case citation: the exact paragraph the proposition is drawn from. It is the difference between telling a marker that a case supports your point and showing them where. AGLC4 requires it, markers deduct for missing it, and it is the single most useful thing in a footnote.
It is also the mechanism behind the whole comparison on this page. A claim without a pinpoint cannot be checked quickly, and a claim that cannot be checked quickly does not get checked. In CaseSharp every claim in an answer carries one, and clicking it opens the judgment at that paragraph rather than at the top of the case.
Why checking whether a case still stands means reading the later cases
Working out whether a case is still good law is not a reasoning problem, it is a reading problem. It means finding the later judgments that have cited the case and seeing what each one did with it, whether it followed, applied, distinguished, doubted or overruled the point you are relying on. That trail has to be built and kept current from the judgments themselves.
A general assistant has no Australian citation graph behind it by default, so it cannot show you which later cases have cited an authority. Browsing and retrieval modes change what a model can reach, and they are genuinely useful for recency, but a page pulled off the web is not the citation trail: it answers what one document says, not which later courts have relied on the case. That is why a fluent summary can describe the position as it stood before a later court narrowed it. CaseSharp does not put a still-good-law label on a case. What it shows on every authority is the citation graph, the cases that cite it and the cases it cites, ranked by influence, each one named and linked, so the later cases you need to read sit next to the case name rather than at the end of a separate lookup.
The part that is actually at stake
Australian law schools treat undisclosed AI use as academic misconduct. The policy is usually set at subject level by the convenor, which means the answer to whether you may use a tool changes between two assessments in the same semester. Where AI use is permitted, a disclosure statement is normally required.
The consequence is what makes this different from other degrees. A misconduct finding has to be disclosed to the admitting authority when you apply to be admitted, years after the assessment that caused it. That is a long shadow to accept in exchange for a faster draft.
The profession has its own version of the same story. Australian courts and regulators have already dealt with lawyers who filed AI-invented citations, and Australian courts have issued practice guidance restricting the use of generative AI in litigation documents. Law faculties have been repeating both facts ever since. We are not naming anyone, because the point is not the individual.
CaseSharp is built the other way around. Nothing it produces is an answer you hand in. Every claim resolves to a paragraph of a real judgment, every substantive output carries a human-review notice, and when there is no authority to support something, Sharp says so instead of filling the gap.
Using both, sensibly
- Use a general assistant to get your head around a doctrine you cannot follow in the textbook. That is what it is good at.
- Never carry a citation across from it into your work without opening the case. If you cannot open it, it does not go in.
- Read the later cases before you rely on anything, including cases you were given in a lecture in week 3.
- Read the AI policy for each individual assessment, and disclose where it asks you to.
Questions students ask about this
- Should I stop using ChatGPT for law?
- Not necessarily, and this page is not going to pretend otherwise. A general assistant is genuinely good at explaining a concept from scratch, at drafting things that are not legal writing, and at working through your non-law subjects. The argument here is narrower: it is the wrong instrument for a task where a wrong citation is a misconduct risk, because it does not pin a claim to a paragraph of the judgment and it has no Australian legal citator behind it.
- Can I use both in the same piece of work?
- Most students do. A reasonable pattern is to use a general assistant to understand a doctrine you are stuck on, then use CaseSharp to find the authority that actually supports the point, check how it has been treated, and produce the citation. Whatever you use, disclose it where your subject requires disclosure, and never cite a case you have not opened.
- Will my university know if I used AI?
- That is a question for your university rather than for us. What matters more is the rule: Australian law schools treat undisclosed AI use as academic misconduct, the policy is often set per subject by the convenor, and where AI use is permitted a disclosure statement is usually required. CaseSharp has no relationship with any university and no mechanism to report on an individual account.
- What is a pinpoint citation?
- It is a citation to the exact paragraph or page a proposition comes from, rather than to the case as a whole. In AGLC4 it is the number after the case citation. It matters because it is the difference between a marker being able to check your claim in five seconds and having to read a sixty page judgment to work out whether you are right.
- Can a general model tell me whether a case is still good law?
- Not on its own, and the reason is structural rather than a matter of how clever the model is. Answering that question properly means reading the later judgments that have cited the case and seeing what each one did with it. Several assistants can now browse or retrieve web pages while they answer, which helps with recency, but a retrieved page will not reliably tell you that a later court narrowed the case unless something it happened to open says so. CaseSharp does not put a still-good-law label on a case either. What it does is pin every claim to the paragraph it came from and show you the citation graph on the authority, the cases that have cited it ranked by influence, each one a click from the judgment, so the reading that answers the question is fast.
73,000 audited Australian decisions, current to the end of April 2026. CaseSharp is a research and study tool. It is not legal advice.