arXiv cs.CL· Elise L. Han, Tatsunori Hashimoto·· 11 小时前AI 评分34
为语言模型水印协调隐私与来源追溯：Red-Green Sequence 水印
Reconciling Privacy with Provenance for Language Model Watermarks
AI 导读
研究者提出本地差分隐私水印，将水印设计目标从追踪个体转向估计总体使用频率，从而协调来源追溯与隐私。其 Red-Green Sequence（RGS）水印在 C4 语言建模任务上以 ε=1 实现 0.0019 MSE 和 0.0038 平均绝对偏差；在仅 100 条文本的作文任务中，ε=2 时误差仍为 0.0113 MSE 和 0.004 平均绝对偏差。
正文 · 原文
Abstract:Watermarking has great potential for regulating the use of language models (LMs), but its deployment has been marred by controversy, as many users argue that fine-grained tracking of LM usage violates their privacy. This tension reduces the effectiveness of watermarks, as users engage in evasion strategies such as paraphrasing that can remove much of the watermark signal. We show that provenance and privacy can be reconciled if we shift the goal of watermark design from tracking individuals to estimating aggregate usage. We develop locally differentially private watermarks which provably limit the evidence that any individual text can contribute to detection, while still enabling accurate estimation of population frequencies. We introduce the Red-Green Sequence (RGS) watermark, a private and computationally efficient watermark that achieves accurate frequency estimation at high privacy levels across various models, sample sizes and generation entropies in settings that reflect realistic deployment scenarios. On C4 language modeling tasks, we attain $0.0019$ MSE with $0.0038$ mean absolute bias at a low $\varepsilon$ of 1, while in more challenging settings such as an essay writing task with only 100 text samples, we still find errors on the order of $0.0113$ MSE with $0.004$ mean absolute bias at $\varepsilon=2$.
| Subjects: | Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) |
| Cite as: | arXiv:2611.06100 [cs.CR] |
| (or arXiv:2611.06100v1 [cs.CR] for this version) | |
| https://doi.org/10.48550/arXiv.2611.06100 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Elise Han [view email]
[v1]
Wed, 4 Nov 2026 07:45:02 UTC (2,625 KB)
来源：arXiv cs.CL · arxiv.org