How Good is Post-Hoc Watermarking With Language Model Rephrasing?
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917206514728960 |
|---|---|
| author | Fernandez, Pierre Sander, Tom Elsahar, Hady Chang, Hongyan Souček, Tomáš Lacatusu, Valeriu Tran, Tuan Rebuffi, Sylvestre-Alvise Mourachko, Alexandre |
| author_facet | Fernandez, Pierre Sander, Tom Elsahar, Hady Chang, Hongyan Souček, Tomáš Lacatusu, Valeriu Tran, Tuan Rebuffi, Sylvestre-Alvise Mourachko, Alexandre |
| contents | Generation-time text watermarking embeds statistical signals into text for traceability of AI-generated content. We explore *post-hoc watermarking* where an LLM rewrites existing text while applying generation-time watermarking, to protect copyrighted documents, or detect their use in training or RAG via watermark radioactivity. Unlike generation-time approaches, which is constrained by how LLMs are served, this setting offers additional degrees of freedom for both generation and detection. We investigate how allocating compute (through larger rephrasing models, beam search, multi-candidate generation, or entropy filtering at detection) affects the quality-detectability trade-off. Our strategies achieve strong detectability and semantic fidelity on open-ended text such as books. Among our findings, the simple Gumbel-max scheme surprisingly outperforms more recent alternatives under nucleus sampling, and most methods benefit significantly from beam search. However, most approaches struggle when watermarking verifiable text such as code, where we counterintuitively find that smaller models outperform larger ones. This study reveals both the potential and limitations of post-hoc watermarking, laying groundwork for practical applications and future research. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_16904 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | How Good is Post-Hoc Watermarking With Language Model Rephrasing? Fernandez, Pierre Sander, Tom Elsahar, Hady Chang, Hongyan Souček, Tomáš Lacatusu, Valeriu Tran, Tuan Rebuffi, Sylvestre-Alvise Mourachko, Alexandre Cryptography and Security Computation and Language Generation-time text watermarking embeds statistical signals into text for traceability of AI-generated content. We explore *post-hoc watermarking* where an LLM rewrites existing text while applying generation-time watermarking, to protect copyrighted documents, or detect their use in training or RAG via watermark radioactivity. Unlike generation-time approaches, which is constrained by how LLMs are served, this setting offers additional degrees of freedom for both generation and detection. We investigate how allocating compute (through larger rephrasing models, beam search, multi-candidate generation, or entropy filtering at detection) affects the quality-detectability trade-off. Our strategies achieve strong detectability and semantic fidelity on open-ended text such as books. Among our findings, the simple Gumbel-max scheme surprisingly outperforms more recent alternatives under nucleus sampling, and most methods benefit significantly from beam search. However, most approaches struggle when watermarking verifiable text such as code, where we counterintuitively find that smaller models outperform larger ones. This study reveals both the potential and limitations of post-hoc watermarking, laying groundwork for practical applications and future research. |
| title | How Good is Post-Hoc Watermarking With Language Model Rephrasing? |
| topic | Cryptography and Security Computation and Language |
| url | https://arxiv.org/abs/2512.16904 |