Data Provenance Auditing of Fine-Tuned Large Language Models with a Text-Preserving Technique
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866914320384786432 |
|---|---|
| author | Li, Yanming Eichler, Cédric Anciaux, Nicolas Bensamoun, Alexandra Manzano, Lorena Gonzalez Ghozzi, Seifeddine |
| author_facet | Li, Yanming Eichler, Cédric Anciaux, Nicolas Bensamoun, Alexandra Manzano, Lorena Gonzalez Ghozzi, Seifeddine |
| contents | 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. 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_09655 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Data Provenance Auditing of Fine-Tuned Large Language Models with a Text-Preserving Technique Li, Yanming Eichler, Cédric Anciaux, Nicolas Bensamoun, Alexandra Manzano, Lorena Gonzalez Ghozzi, Seifeddine Cryptography and Security Artificial Intelligence 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. 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. |
| title | Data Provenance Auditing of Fine-Tuned Large Language Models with a Text-Preserving Technique |
| topic | Cryptography and Security Artificial Intelligence |
| url | https://arxiv.org/abs/2510.09655 |