Methodology and source record
How matters enter the corpus, how classifications are applied, what the tracker can and cannot establish, and how to inspect or correct the underlying record.
What the tracker covers
AI Vortex tracks public court matters, orders, disciplinary records, and other documented legal proceedings in which generative AI, fabricated authority, inaccurate quotations, unsupported propositions, or related verification failures are part of the public record.
Inclusion standard
- A publicly identifiable legal matter or proceeding exists.
- A source document or attributable public record supports inclusion.
- The record contains a legal AI issue, verification failure, or closely related allegation.
- The matter can be classified without inventing facts absent from the source.
Read status before outcome. “Non-alleged” is a corpus flag, not proof of a final adjudicated misconduct finding. Warnings, show-cause orders, procedural rulings, and final outcomes require separate reading of the underlying record.
Classification framework
Editorial impact
Summarizes the observed procedural, monetary, professional, or case-level consequence. It is not a prediction of future sanctions.
Failure modes
Tags can overlap. One matter may contain fake citations, fabricated quotations, and misrepresented authority.
AI tools
Tool names appear only when recorded. “Unidentified” does not imply a vendor, and counts are not usage-adjusted rates.
Source and review hierarchy
Corpus provenance.The base incident archive and many preserved source documents are supplied through Damien Charlotin's public legal AI case archive. AI Vortex independently imports, normalizes, classifies, enriches, presents, and maintains this research layer; that work does not imply endorsement of AI Vortex or its classifications.
A link hosted by the issuing court, tribunal, or government domain.
A repository preserving the filing, order, opinion, or docket material.
An upstream archive preserving the relevant record without being the issuing court.
Contextual material that requires additional source review before reliance.
Source tier describes where the link points. It does not certify that AI Vortex independently verified every field. Treat the linked document as controlling and report discrepancies for review.
Public indexing and evidence depthAll 1,870 corpus records have stable, indexable public URLs. 1,611 currently meet the stronger publication and evidence baseline; the remaining 259 stay indexable with their documented limitations disclosed on-page.
Current review programSource traceability and correction intake are live. A corpus-wide independent field-by-field review is not represented as complete.
Portraits, courthouse images, and visual fallbacks
Entity images are local, versioned research assets. The current registry includes 21 verified judicial portraits and 9 verified representative courthouse images. Every real image retains its creator, source page, reuse license, alternative text, and caption in the page, report, and structured metadata.
Real-image gateA photograph appears only after the pictured person or building and the reuse terms are verified. Wikimedia Commons, official court materials, the Federal Judicial Center, and GSA records may support that review; no Google Images result is treated as a license.
Complete court coverageAll 622 normalized court profiles receive either a licensed representative image or a first-party scope marker derived from recorded geography and court classification. Scope markers are not photographs, official seals, or location maps.
Media registry revision 2026-07-28. A courthouse image may show one representative building and does not imply that a court sits only at that location.
Updates, corrections, and reproducibility
The upstream public dataset was checked and validated on this date. The latest tracked decision is Aug 11, 2026.
New documents, changed outcomes, and corrections are incorporated into later snapshots.
Submit a case, source link, court order, or correction with supporting documentation.