Metraipolis › archive › dc_0579b76ec65c
Language Model Fingerprinting Requires Rethinking Watermark Teachers — arXiv:2610.04169
- Source
- arXiv
- Original
- https://arxiv.org/abs/2610.04169
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-06T05:26:02.363Z
- SHA-256 of the archived text
e6c292108cf89e7072bfb4c1010a4b9562c974538a1330b382aff37ffd66814d- Stable URL of this page
https://metraipolis.com/documento/dc_0579b76ec65c
How to cite
Language Model Fingerprinting Requires Rethinking Watermark Teachers — arXiv:2610.04169 — arXiv, https://arxiv.org/abs/2610.04169 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0579b76ec65c (Metraipolis archive dc_0579b76ec65c, sha256 e6c292108cf89e7072bfb4c1010a4b9562c974538a1330b382aff37ffd66814d)
BibTeX
@misc{dc0579b76ec65c,
title = {Language Model Fingerprinting Requires Rethinking Watermark Teachers — arXiv:2610.04169},
howpublished = {arXiv},
url = {https://arxiv.org/abs/2610.04169},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0579b76ec65c; sha256 e6c292108cf89e7072bfb4c1010a4b9562c974538a1330b382aff37ffd66814d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.04169. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
LLM fingerprinting via watermark distillation embeds a statistical watermark signal into model weights, enabling model owners to identify their models behind black-box APIs. Revisiting a recent protocol, we find that its utility evaluation understates text quality degradation in open-ended generation, favoring overly strong watermark teachers. Weakening the watermark improves text quality but sacrifices detectability. To move beyond this trade-off, we rethink whether text watermarks designed for verifying generated text are suitable distillation teachers for model fingerprinting. Such watermarks are typically designed to remain detectable from an individual output, limiting how sparse the watermark signal can be. In contrast, fingerprint verification can aggregate signal across queries, making sparser watermark signals viable. This raises a key question: where should the sparse signal be placed? We analyze signal placement through token surprisal and show that, even at comparable watermark strength, different placements can target tokens with different plausibility under the base model. This motivates near-tie restriction, which uses top-1-relative logit gaps to restrict the watermark bias to tokens close to the base model's top prediction. Across multiple models, near-tie improves detection--quality frontiers under deployment changes, preserves higher text quality across query budgets, and further improves existing watermarking schemes when combined with them. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Cryptography and Security (cs.CR) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.04169 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-06T05:24:05.805Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).