~4 min read

Why not just use Westlaw's cite-check?

"Just use KeyCite" and "LEXIS does Shepardize" are the most common objections to a citation-validation tool. They're worth taking seriously — Westlaw and LEXIS are excellent products that millions of attorneys use daily. But they answer a different question than "does this case actually exist?"

What cite-checkers actually check

KeyCite and Shepard's tell you:

What they assume going in: that the case you're checking actually exists. The query "is Smith v. Johnson, 999 F.3d 1234 still good law?" presupposes that Smith v. Johnson is a real case.

What ChatGPT-fabricated cases look like

The fabricated citations in Mata v. Avianca followed canonical Bluebook format. They had real-looking volume numbers, real reporter abbreviations, and plausible page ranges. The attorneys ran them through Westlaw — and got results, just not for the cases the brief claimed to cite.

The pattern: Westlaw returned partial matches or "no results found", which the attorney interpreted as a search miss rather than a fabrication flag. Without a deliberate "does this exact citation correspond to this exact case name?" check, the gap went undetected.

The check Cite Scan runs

Cite Scan's primary call resolves citation strings against the open CourtListener index. For each cite:

A fabricated cite typically fails the first check (no match) and may fail the third (the AI invented a reporter abbreviation that doesn't exist). Both are red flags Cite Scan raises in seconds.

This isn't a Westlaw replacement

Cite Scan's job is the pre-flight check: does this case exist? Westlaw's job is the depth check: is it still good law? what does it stand for? Both matter. The order matters too — asking the second question of a fabricated case is incoherent.

Most law firms will continue paying for Westlaw or LEXIS. Proof Brief is the $39/month layer in front of those tools that catches the cases your premium subscription was never designed to flag.

Try the 14-day trial