CiteProof / Product

Built to earn the signature, not just the click.

Anyone can run a citation past a search box. CiteProof exists for the moment a partner signs a brief: every flag is evidence you can inspect, every limit is stated on the page, and the checker is deterministic — no AI model ever sees your brief, though parser flags can still be wrong and are shown for inspection.

5 sections

What you get that a search box doesn't

  • Whole-brief gate. Paste the brief once; each citation and quoted passage we can read is extracted and checked in one run, with a counts line proving what was read. Anything unread reads as Couldn't check, never as verified.
  • Identity, not just existence. A fabricated citation can land on a real reporter location occupied by a different case. We compare the name, court and year your brief claims against the record — and show both sides when they differ.
  • Corrected-cite suggestions. When the brief names the wrong case at a real location, the report proposes the resolved cite as a starting point — a suggestion only, never an edit to your filing.
  • Quote forensics (paid). Side-by-side passages with pin cites where available, match confidence, guardrail reasons, and word-level diffs for dropped negations and changed numbers — when the opinion text is retrieved; unfetched quotes stay uncompared as Couldn't check.
  • A sign-off workflow. Flags become review tasks with a progress count, a copyable review log, and a downloadable Verification Record for the file.
  • A record bound to its brief. Every paid Verification Record carries the SHA-256 of the exact checked text with a scannable audit code, so a filed copy shows which bytes it checked.
  • Honest boundaries. Good-law verdicts, support-for-proposition and unpublished/state-trial coverage are things we don't do — and the report says so instead of guessing.
  • U.S. Code existence. Federal statute sections are confirmed against GovInfo's Code edition — existence only, never currency or interpretation, and a section that isn't there flags for review instead of passing silently.

Measured, not promised

Project measurements, October 2026 runs against real case text. Read them with the caveats — a small holdout bounds the false-clean rate; it does not certify it.

  • 0 false cleans on 17 bad holdout items (rule-of-three upper bound about 18%, reported as a bound, never a rate). The 6 dev items tuned the matcher and count as no evidence.
  • 18 of 18 planted citations extracted across 6 synthetic briefs.
  • Identity comparison judged on ~37.5% of full citations where a usable party name is available — the rest get existence-only checking, stated as such, never a silent pass.

Built for the ethical duty, not instead of it

Mata v. Avianca, Inc. S.D.N.Y. 2023 · Rule 11 sanctions

Court sanctioned attorneys for submitting non-existent judicial decisions with bogus quotes generated by an LLM without independent verification against official reporters.

Park v. Kim 2d Cir. 2024 · Appellate referral

Second Circuit referred counsel to grievance committee after brief cited an invented case, reiterating that a lawyer who submits AI output without verification breaches the duty of candor.

Courts have sanctioned lawyers for filing AI-hallucinated citations; state bars increasingly tie technology competence to Rule 1.1, and ABA Formal Opinion 512 (2024) addresses generative AI use directly, alongside the candor duty of Rule 3.3. The limit first: CiteProof satisfies no duty by itself, attorney review remains required, and the record proves nothing about your review. What it does do is document what was checked and when — a hashed, timestamped record a lawyer can describe when showing reasonable inquiry under Rule 11.

How this compares

Capability by capability, against the industry standards attorneys actually use. Competitor cells describe vendor-published features and third-party reported pricing, October 2026 — confirm current details with each vendor. We publish the rows where we lose on purpose: knowing what CiteProof doesn't do is part of using it safely.

CapabilityCiteProof ($49/brief, free to try)Clearbrief (Word add-in, per-seat SaaS)Westlaw CoCounsel / Lexis+ Brief Analysis (subscription)Free lookup (CourtListener / Scholar)
Whole-brief existence + identity checkYes — free, no signup, one runYes, in WordYes (Brief Analysis / Document Analyzer)By hand in the browser, or self-built on their batch API — no report, no quotes, no workflow
U.S. Code section verificationYes — GovInfo, existence in the stated editionCitation checking in Word; confirm statute depth with vendorBrief review checks citations; confirm statute depth with vendorBy hand, section by section
Verification diffs between brief revisionsYes — resubmit with a supersedes link; history and API show added, removed and changed, regressions firstNot a listed featureNot a listed featureNo
Secondary cites named by kindYes — treatise, restatement, journal, encyclopedia, manual, each with its next stepNot a listed featureNot a listed featureNo
Coverage pre-check by court levelYes — published before you pay, no invented ratesNot a listed featureNot a listed featureNo
Parallel reporter locationsYes — folded to one row, each location resolved, disagreements flaggedNot a listed featureNot a listed featureNo
Quote-text verification with pin citeYes, paid — side-by-side, match confidence, word-level diffsYesYes (Quote Check with pinpoint validation)No
Good-law verdictNo — citing-history starting points, never a verdictVia LexisNexis dataYes — Shepard's / KeyCite, the industry standardNo
Fact-citation checkingNoYes — facts and lawPartialNo
Support-for-proposition analysisNoSemantic scoresPartialNo
Partner sign-off workflowYes — review tasks, progress, log, Verification RecordYes — Cite Check Report audit trail (Dec 2025)Partial reportsNo
Adversary audit of the other side's briefYes — paid per audit: citation mismatches, word-level quotation diffs, unverifiable record cites, confirm-or-dismiss review, Audit RecordConfirm with vendorConfirm with vendorNo
Corrected-cite suggestions on mismatchesYes — "did you mean" with the resolved cite; suggestion only, never appliedConfirm with vendorConfirm with vendorNo
Practice-management sync (Clio)Yes — import, bulk sync, save-back, diagnosticsMyCase partnership (confirm scope with vendor)Confirm with vendorNo
Tamper-evident record (brief hash + scannable code)Yes — SHA-256 audit block on every paid recordConfirm with vendorConfirm with vendorNo
Filing-ready artifactVerification Record PDF (audit block included) + Table of AuthoritiesCite Check Report + hyperlinked filingBrief Analysis reportNone
Programmatic accessYes — API v1: single, batch, batch-zip, cites-only, brief comparisonEnterprise integrationsEnterprise APIsPublic API, self-built
Price to check one brief$49, or $0 to tryPer-seat SaaS — vendor quoteReported ~$170–700/mo per seat$0 plus your time
FrictionNo account; one brief runs in the browser (the Word add-in needs an IT-approved sideload)Word install + sales conversationAnnual contract + IT procurementManual, cite by cite
Brief confidentialityNo AI model ever sees your brief; free checks store nothingSOC-2, enterprise DPAEnterprise DPA; AI features process brief textQueries reveal the citations you look up

Reading the table: CiteProof wins the single row of checking one brief tonight with no procurement. The subscriptions win every row about depth. Buy us for the gate before filing; keep the citator for good law.

When not to pay us

  • You need to know whether authority is still good law — buy the citator check, or use our citing-history links as the starting point.
  • Your brief is mostly regulations, state statutes or secondary sources — we check reported opinions and the U.S. Code.
  • You need a certificate of diligence — our record documents what was checked; it proves nothing about your review.

Citation Check See it work on real examples

CiteProof verifies citation existence and quote accuracy against public court records. It is not legal advice and does not replace attorney review. Terms · Privacy · Contact