TrustCitation

Legal Citation Validation

Know why your citations fail

Existing tools tell you a citation is wrong. TrustCitation tells you exactly how — wrong holding, wrong jurisdiction, non-binding authority, or fabricated entirely.

Get Early Access

Pass/fail isn't enough

AI-generated legal briefs hallucinate citations at alarming rates. Studies show 17-33% of citations in AI-assisted filings are problematic — even with retrieval-augmented tools.

Current citation checkers give you a binary answer: PASS or FAIL. But a citation can fail in five fundamentally different ways — and each one requires 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

TrustCitation runs every citation through a multi-layer validation pipeline and returns a diagnostic report — not a traffic light.

01

Existence Check

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

02

Semantic Match

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

03

Authority Check

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

What a diagnostic report looks like

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 peer-reviewed research

TrustCitation's validation methodology is grounded in published epistemological research on AI knowledge verification — not just pattern matching.

Read the research paper →

Get Early Access

We're opening TrustCitation to a small group of lawyers first.