
YL
Yanming Li (PETSCRAFT), C\'edric Eichler (PETSCRAFT), Nicolas Anciaux (PETSCRAFT), Alexandra Bensamoun (UC3M), Lorena Gonzalez Manzano (UC3M), Seifeddine Ghozzi (ENSTA)
· 1 min read
ResearcharXiv cs.AI
Data Provenance Auditing of Fine-Tuned Large Language Models with a Text-Preserving Technique
arXiv:2510.09655v3 Announce Type: replace-cross
Abstract: We propose a system for marking sensitive or copyrighted texts to detect their use in fine-tuning large language models under black-box access with statistical guarantees. Our method builds digital ``marks'' using invisible Unicode characters organized into (``cue'', ``reply'') pairs. During an audit, prompts containing only ``cue'' fragments are issued to trigger regurgitation of the corresponding ``reply'', indicating document usage. To control false positives, we compare against held-out counterfactual marks and apply a ranking test, yielding a verifiable bound on the false positive rate. Empirically, we obtain a true positive rate of 96.7% at 0% false positive rate and reply regurgitation rates exceeding 28% per document with only 40 (4%) watermarked documents. The approach is minimally invasive, scalable across many sources, robust to standard processing pipelines, and achieves high detection power even when marked data is a small fraction of the fine-tuning corpus.
Original source
This story was published by arXiv cs.AI and written by Yanming Li (PETSCRAFT), C\'edric Eichler (PETSCRAFT), Nicolas Anciaux (PETSCRAFT), Alexandra Bensamoun (UC3M), Lorena Gonzalez Manzano (UC3M), Seifeddine Ghozzi (ENSTA). SyncAI.news shows a preview; the complete article is on the publisher's site.
Read the full story on arxiv.org


