mlsanityData to 7 Oct 2026

Optimal Multi-bit Generative Watermarking Schemes Under Worst-Case False-Alarm Constraints

arXiv 2604.08759 ↗First posted 9 Apr 2026cs.IT cs.CL math.ITcited by 0

On the map: LLM safety and security → Watermarking generated content and models

Abstract (arXiv)

This paper considers the problem of multi-bit generative watermarking for large language models under a worst-case false-alarm constraint. Prior work established a lower bound on the achievable miss-detection probability in the finite-token regime and proposed a scheme claimed to achieve this bound. We show, however, that the proposed scheme is in fact suboptimal. We then develop two new encoding-decoding constructions that attain the previously established lower bound, thereby completely characterizing the optimal multi-bit watermarking performance. Our approach formulates the watermark design problem as a linear program and derives the structural conditions under which optimality can be achieved. In addition, we identify the failure mechanism of the previous construction and compare the tradeoffs between the two proposed schemes.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Watermarking generated content and models, one of 12 directions in LLM safety and security. Builds on 27 of its 36 references that are in the map. First posted on arXiv 9 Apr 2026.

Builds on

27 of 36 references.

Closest ancestors · 26

Same direction, newest first

  1. 2025A Reinforcement Learning Framework for Robust and Secure LLM Watermarking23 Oct
  2. 2025StealthInk: A Multi-bit and Stealthy Watermark for Large Language Models5 Jun
  3. 2025Optimized Couplings for Watermarking Large Language Models13 May
  4. 2025Improved Unbiased Watermark for Large Language Models16 Feb
  5. 2024Theoretically Grounded Framework for LLM Watermarking: A Distribution-Adaptive Approach3 Oct
  6. 2024Watermarking Language Models with Error Correcting Codes12 Jun
  7. 2024Learning to Watermark LLM-generated Text via Reinforcement Learning13 Mar
  8. 2024Duwak: Dual Watermarks in Large Language Models12 Mar
  9. 2024Multi-Bit Distortion-Free Watermarking for Large Language Models26 Feb
  10. 2024Adaptive Text Watermark for Large Language Models25 Jan
  11. 2023Optimizing watermarks for large language models28 Dec
  12. 2023A Survey of Text Watermarking in the Era of Large Language Models13 Dec
14 more
  1. 2023On the Learnability of Watermarks for Language Models7 Dec
  2. 2023A Resilient and Accessible Distribution-Preserving Watermark for Large Language Models11 Oct
  3. 2023Necessary and Sufficient Watermark for Large Language Models2 Oct
  4. 2023Unbiased Watermark for Large Language Models22 Sep
  5. 2023CodeMark: Imperceptible Watermarking for Code Datasets against Neural Code Completion Models28 Aug
  6. 2023Advancing Beyond Identification: Multi-bit Watermark for Large Language Models1 Aug
  7. 2023An Unforgeable Publicly Verifiable Watermark for Large Language Models30 Jul
  8. 2023Towards Codable Watermarking for Injecting Multi-bits Information to LLMs29 Jul
  9. 2023Robust Distortion-free Watermarks for Language Models28 Jul
  10. 2023Provable Robust Watermarking for AI-Generated Text30 Jun
  11. 2023On the Reliability of Watermarks for Large Language Models7 Jun
  12. 2023Undetectable Watermarks for Language Models25 May
  13. 2023DeepTextMark: A Deep Learning-Driven Text Watermarking Approach for Identifying Large Language Model Generated Text9 May
  14. 2023A Watermark for Large Language Models24 Jan

Other references · 1

  1. 2025Deep Learning-based Dual Watermarking for Image Copyright Protection and Authenticationcited by 3

Built on by

Not cited yet.