Can ChatGPT make up court cases?

Yes — and it already has, in real filings. A language model can generate a case name, a realistic reporter citation and a convincing holding for an authority that does not exist. Lawyers have been sanctioned for filing briefs containing these fabricated cases, so any AI-assisted citation must be verified against a real legal database before filing.

How do I check whether a case citation is real?

Paste the case citation into CiteProve. It searches CourtListener for a matching case name and reporter citation. If no such case can be located, it is reported as not found — carefully worded, because absence of a record is a flag to verify, not an accusation. Real cases with a wrong reporter, court or year are flagged as partial.

Is this a substitute for Westlaw or Lexis?

No. CiteProve is a fast first-pass existence and metadata check built on the public CourtListener database — it tells you whether a cited authority is real and correctly described. It does not replace full legal research, KeyCite/Shepard's treatment analysis, or your professional judgment.

Verify case citations before you file

AI drafting tools invent court cases that look entirely real — and filing one has already cost lawyers sanctions. Paste the citations from a brief and check each one against CourtListener before it reaches a judge.

In short

To check whether a case citation is real, look it up in a legal database rather than trusting the brief. AI assistants fabricate cases with realistic names and reporter citations, and courts have sanctioned attorneys who filed them. CiteProve searches CourtListener for each citation and flags any case it cannot locate, plus real cases cited to the wrong reporter, court or year — a fast first-pass check before you file. Free, no account.

Paste the citations from your brief

The risk AI has introduced to legal writing

Generative tools produce fluent, confident text — including citations. Because a fabricated case carries a plausible name and a well-formed reporter cite, it can pass a quick read by a busy attorney. The consequences are no longer hypothetical: courts across multiple jurisdictions have imposed sanctions and public reprimands for briefs built on non-existent authority. Verification is now part of professional diligence.

What CiteProve checks for a legal brief

Entirely fabricated cases with realistic reporter citations
Real reporter cites attached to the wrong case name
Wrong court or year on otherwise real cases
Bluebook-formatted references parsed and looked up directly

Built on the trust principle

CiteProve flags; it never accuses. A citation it cannot locate is reported as not found with careful wording, every verdict lists the source consulted, and anything outside coverage is reported as can't-check rather than guessed. The final call — and the filing — stays yours.

Frequently asked questions

Can ChatGPT make up court cases?: Yes — and it already has, in real filings. A language model can generate a case name, a realistic reporter citation and a convincing holding for an authority that does not exist. Lawyers have been sanctioned for filing briefs containing these fabricated cases, so any AI-assisted citation must be verified against a real legal database before filing.
How do I check whether a case citation is real?: Paste the case citation into CiteProve. It searches CourtListener for a matching case name and reporter citation. If no such case can be located, it is reported as not found — carefully worded, because absence of a record is a flag to verify, not an accusation. Real cases with a wrong reporter, court or year are flagged as partial.
Is this a substitute for Westlaw or Lexis?: No. CiteProve is a fast first-pass existence and metadata check built on the public CourtListener database — it tells you whether a cited authority is real and correctly described. It does not replace full legal research, KeyCite/Shepard's treatment analysis, or your professional judgment.