StealthInk: A Multi-bit and Stealthy Watermark for Large Language Models

Ya Jiang, Chuxiong Wu, Massieh Kordi Boroujeny, Brian Mark, Kai Zeng
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:27685-27709, 2025.

Abstract

Watermarking for large language models (LLMs) offers a promising approach to identifying AI-generated text. Existing approaches, however, either compromise the distribution of original generated text by LLMs or are limited to embedding zero-bit information that only allows for watermark detection but ignores identification. We present StealthInk, a stealthy multi-bit watermarking scheme that preserves the original text distribution while enabling the embedding of provenance data, such as userID, TimeStamp, and modelID, within LLM-generated text. This enhances fast traceability without requiring access to the language model’s API or prompts. We derive a lower bound on the number of tokens necessary for watermark detection at a fixed equal error rate, which provides insights on how to enhance the capacity. Comprehensive empirical evaluations across diverse tasks highlight the stealthiness, detectability, and resilience of StealthInk, establishing it as an effective solution for LLM watermarking applications.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-jiang25j, title = {{S}tealth{I}nk: A Multi-bit and Stealthy Watermark for Large Language Models}, author = {Jiang, Ya and Wu, Chuxiong and Boroujeny, Massieh Kordi and Mark, Brian and Zeng, Kai}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {27685--27709}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/jiang25j/jiang25j.pdf}, url = {https://proceedings.mlr.press/v267/jiang25j.html}, abstract = {Watermarking for large language models (LLMs) offers a promising approach to identifying AI-generated text. Existing approaches, however, either compromise the distribution of original generated text by LLMs or are limited to embedding zero-bit information that only allows for watermark detection but ignores identification. We present StealthInk, a stealthy multi-bit watermarking scheme that preserves the original text distribution while enabling the embedding of provenance data, such as userID, TimeStamp, and modelID, within LLM-generated text. This enhances fast traceability without requiring access to the language model’s API or prompts. We derive a lower bound on the number of tokens necessary for watermark detection at a fixed equal error rate, which provides insights on how to enhance the capacity. Comprehensive empirical evaluations across diverse tasks highlight the stealthiness, detectability, and resilience of StealthInk, establishing it as an effective solution for LLM watermarking applications.} }
Endnote
%0 Conference Paper %T StealthInk: A Multi-bit and Stealthy Watermark for Large Language Models %A Ya Jiang %A Chuxiong Wu %A Massieh Kordi Boroujeny %A Brian Mark %A Kai Zeng %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-jiang25j %I PMLR %P 27685--27709 %U https://proceedings.mlr.press/v267/jiang25j.html %V 267 %X Watermarking for large language models (LLMs) offers a promising approach to identifying AI-generated text. Existing approaches, however, either compromise the distribution of original generated text by LLMs or are limited to embedding zero-bit information that only allows for watermark detection but ignores identification. We present StealthInk, a stealthy multi-bit watermarking scheme that preserves the original text distribution while enabling the embedding of provenance data, such as userID, TimeStamp, and modelID, within LLM-generated text. This enhances fast traceability without requiring access to the language model’s API or prompts. We derive a lower bound on the number of tokens necessary for watermark detection at a fixed equal error rate, which provides insights on how to enhance the capacity. Comprehensive empirical evaluations across diverse tasks highlight the stealthiness, detectability, and resilience of StealthInk, establishing it as an effective solution for LLM watermarking applications.
APA
Jiang, Y., Wu, C., Boroujeny, M.K., Mark, B. & Zeng, K.. (2025). StealthInk: A Multi-bit and Stealthy Watermark for Large Language Models. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:27685-27709 Available from https://proceedings.mlr.press/v267/jiang25j.html.

Related Material