TrustCitation

Legal Citation Validator

One fabricated citation can cost your license.

AI tools are inventing cases. Courts are sanctioning attorneys for filing them. TrustCitation catches every hallucinated, misattributed, or wrong-jurisdiction citation — expert-reviewed report in your inbox within 24 hours.

Send your brief for review →

Pass/fail isn't enough

Studies show 17–33% of citations in AI-assisted briefs are problematic — even with retrieval-augmented tools. Existing checkers return a binary answer: PASS or FAIL. Not enough. A citation can fail in five fundamentally different ways, and each one needs a different fix.

Fabricated

The case doesn't exist at all

Schwartz v. Avianca — entirely invented by ChatGPT

Misattributed Holding

Case exists, but doesn't say what's claimed

Real case cited for a holding it never made

Wrong Jurisdiction

Persuasive authority cited as binding

9th Circuit case applied in 2nd Circuit brief

Hallucinated Detail

Case exists but key details are wrong

Wrong year, wrong court, wrong parties

Coverage Gap

Case isn't in any public database

Unreported decision or restricted access source

Diagnosis, not just detection

Every brief runs through a four-layer pipeline. Your report lands in your inbox within 24 hours.

01

Existence Check

Cross-reference against public court databases. Does this case actually exist?

02

Semantic Match

Read the opinion. Does it actually say what your brief claims it says?

03

Authority Check

Is this binding or persuasive? Right jurisdiction? Right court level?

04

Expert Review

A human reviewer validates every flag before delivery. No output ships unchecked.

This is what lands in your inbox

Every citation gets a confidence score, failure category, and plain-English explanation.

Citation Report — Sample Brief

3 citations analyzed

Schwartz v. Avianca, No. 22-cv-1461 (S.D.N.Y. 2023)

FABRICATED
94%

This case does not exist in any public court database. The docket number, parties, and court are entirely fabricated. This is a known AI hallucination — cited in Mata v. Avianca as a cautionary example.

Terry v. Ohio, 392 U.S. 1 (1968)

MISATTRIBUTED
78%

Case exists and is correctly cited. However, the brief claims this case established a “reasonable expectation of privacy” standard. Terry v. Ohio actually established the “reasonable suspicion” standard for stop-and-frisk. The claimed holding is misattributed.

Miranda v. Arizona, 384 U.S. 436 (1966)

VALID
97%

Citation verified. Case exists, holding matches claimed proposition, binding authority in this jurisdiction.

Built on published epistemology research.

TrustCitation applies a peer-reviewable framework — six evidence-based instruments for verifying knowledge claims — to legal citation checking. Built by someone who specialises in intelligent products and systems thinking. A decade inside founding teams as designer, brand lead, and PM — now pushing code too. The research paper goes deep on the theory; the product just applies it.

Read the research paper →

Send your brief for review

Email your brief as a PDF or pasted text. We run it through the full pipeline, a human reviews every flag, and your report is in your inbox within 24 hours. Lawyers and paralegals welcome.

[email protected]

No signup. No payment. Wave 1 is free while we learn what you need.