+VLFBERHT+ | Legal
AI research that won’t get you sanctioned.
After Mata v. Avianca, every firm knows AI fabricates citations. Ulfberht intercepts citation hallucinations, jurisdiction errors, and precedent misapplication before they reach a brief, motion, or memorandum.
The Warning
$12,622 in sanctions.
Six fabricated cases.
In Mata v. Avianca (S.D.N.Y. 2023), attorneys submitted a brief containing six AI-generated citations to cases that did not exist. Judge Castel imposed $12,622 in monetary sanctions and required corrective submissions. The ABA issued formal guidance. Courts are now watching every AI-assisted filing.
Court Record — S.D.N.Y. 2023
“The Court finds that the submissions contain fabricated judicial opinions with false internal citations and quotes. Sanctions are warranted.”
Mata v. Avianca, Inc., No. 22-cv-1461 (S.D.N.Y. June 22, 2023)
Three Failure Modes
Risk 01
Citation Fabrication
AI invents case names, reporter citations, and holdings that do not exist. These fabrications are internally consistent, formatted correctly, and pass superficial review. The only defense is verification against authoritative legal databases.
Detection method: Reporter cross-reference + court docket lookup
Risk 02
Precedent Misapplication
Real cases cited for propositions they do not support. The case exists, but the holding is misrepresented, the dictum is elevated to holding, or the ruling applies in a different procedural posture than claimed.
Detection method: Holding extraction + proposition mapping
Risk 03
Jurisdictional Error
AI applies law from the wrong jurisdiction without flagging the conflict. State versus federal law, intra-circuit splits, superseded statutes, and binding versus persuasive authority are common undetected failure modes.
Detection method: Jurisdiction binding authority tree analysis
How It Works
Verification runs before the brief leaves your desk.
Every citation in an AI-assisted document passes through a multi-layer verification chain before Ulfberht returns a status. Nothing is assumed. Every claim is traced to an authoritative source or flagged for attorney review.
- 01
Citation Extraction
Every case citation, statute reference, and regulatory cite is extracted from the document via structured parsing.
- 02
Reporter Lookup
Citations are checked against federal and state reporter indexes. Non-existent citations are flagged immediately.
- 03
Holding Verification
Verified citations are cross-checked to confirm the case actually supports the proposition for which it is cited.
- 04
Audit Report Generation
A structured verification report is generated documenting every verified claim, every flagged claim, and the attorney actions required.
Capabilities
Six verification layers for legal AI.
Each capability module runs independently and can integrate into existing document review workflows via API or direct integration.
01
Case Law Verification
Every cited case is verified against federal and state reporter indexes. Case name, reporter volume, page, holding, and current status are all checked. Reversed, overruled, and distinguished authority is flagged before it reaches a brief.
See claim verification →02
Statute Cross-Check
Statutory citations are verified against the current U.S. Code, CFR, and state codes. Superseded statutes, amended provisions, sunset clauses, and incorrect section references are caught before they create malpractice exposure.
See claim verification →03
Jurisdiction Validation
Automatic detection of cross-jurisdictional citation errors. Federal versus state law conflicts, circuit split awareness, diversity jurisdiction binding authority mapping, and persuasive versus binding authority classification.
See oversight tiers →04
Brief Pre-Filing Audit
A complete verification pass on every AI-assisted brief, motion, or memorandum before filing. Produces a structured certification report documenting every verified citation and every claim requiring attorney review, suitable for co-counsel records.
See the engine →05
Precedent Strength Analysis
Beyond existence checks, Ulfberht evaluates how strongly each cited case actually supports the proposition asserted. Dicta elevated to holdings, distinguishable facts, and overruled lines of cases are surfaced with actionable confidence scores.
See dual-view →06
Opposing Argument Detection
Adversarial analysis identifies cases and statutes that opposing counsel is likely to cite against your position. Circuit-level splits and conflicting precedents in your jurisdiction are surfaced before they become surprises at oral argument.
See dual-view →Use Cases
Where legal teams deploy Ulfberht.
Verification integrates into the workflows where AI-generated content carries the highest professional responsibility exposure.
The Risk Landscape
The stakes for unverified AI in legal practice.
Courts have sanctioned attorneys. Bar associations have issued guidance. Malpractice insurers have flagged AI-generated legal research as a coverage concern. The verification gap is real and the consequences are already on the record.
Data sourced from public court sanctions orders, ABA Formal Opinion 512, state bar guidance documents, and published malpractice insurance advisories. Directional figures only. Not a guarantee of outcomes.
$12,622
Documented
Sanctions imposed in Mata v. Avianca for six fabricated AI citations
6
Documented
Fabricated case citations in a single AI-generated brief filing
ABA 512
Published 2023
ABA Formal Opinion governing attorney duty of competence with AI tools
50+
Directional
State and federal courts with standing orders addressing AI-generated filings
Deployment
Integrates into your existing document workflow.
Ulfberht connects to document management systems, review platforms, and legal research workflows via API. Verification runs in the background and surfaces findings inside the tools your team already uses.
Integration
Document Review API
REST + webhook delivery
Integration
iManage / NetDocuments
Native DMS connectors
Integration
Microsoft 365 Copilot
Word plugin + verification layer
Delivery
Audit Report PDF
Court-ready certification record
+VLFBERHT+ | Legal
Run your AI legal research through verification before it costs you.
Schedule a technical demo with our legal team. We will run a live verification pass on a sample AI-generated brief and show you exactly what the audit report surfaces.
What the demo covers
Live citation verification on a sample AI-generated brief
Audit report format and court-ready certification output
Integration options for your DMS and existing workflow
Deployment scope, timeline, and enterprise pricing