Abstract
A method for detecting AI-generated passages in legal documents by producing deterministic text fingerprints from each paragraph or section and comparing those fingerprints against a labeled reference corpus of known AI-generated legal writing and known human-authored filings. The method normalizes legal text, canonicalizes tokens against an ontology organized by legal-writing dimensions, produces a fixed-width numeric fingerprint per chunk, and returns a per-paragraph AI-likelihood score together with a document-level composite. The output is intended as a triage signal for attorneys, legal operations teams, and courts subject to standing orders requiring AI disclosure.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Burton, Aaron, "Deterministic Text Fingerprinting of Legal Documents for Detecting AI-Generated Passages", Technical Disclosure Commons, (April 21, 2026)
https://www.tdcommons.org/dpubs_series/9859