paper-with-me

Papers

Lost in Overlap: Exploring Logit-based Watermark Collision in LLMs

2024-03-15 · Yiyang Luo, Ke Lin, Chao Gu, Jiahui Hou, Lijie Wen, Ping Luo

The proliferation of large language models (LLMs) in generating content raises concerns about text copyright. Watermarking methods, particularly logit-based approaches, embed imperceptible identifiers into text to address these challenges. However, the widespread usage of watermarking across diverse LLMs has led to an inevitable issue known as watermark collision during common tasks, such as paraphrasing or translation. In this paper, we introduce watermark collision as a novel and general philosophy for watermark attacks, aimed at enhancing attack performance on top of any other attacking methods. We also provide a comprehensive demonstration that watermark collision poses a threat to all logit-based watermark algorithms, impacting not only specific attack scenarios but also downstream applications.

📄 PDF Abstract BibTeX arXiv:2403.10020

Code (0)

등록된 구현이 없습니다.

Tasks

PhilosophyQuestion Answering

Similar Papers 제목 키워드 기반

Distortion-free Watermarks are not Truly Distortion-free under Watermark Key Collisions

2024-06-02 · Yihan Wu, Ruibo Chen, Zhengmian Hu, Yanshuo Chen 외

Language model (LM) watermarking techniques inject a statistical signal into LM-generated content by substituting the random sampling process with pseudo-random sampling, using watermark keys as the random seed. Among th…

Language ModelingLanguage Modelling

A Semantic Invariant Robust Watermark for Large Language Models

2023-10-10 · Aiwei Liu, Leyi Pan, Xuming Hu, Shiao Meng 외

Watermark algorithms for large language models (LLMs) have achieved extremely high accuracy in detecting text generated by LLMs. Such algorithms typically involve adding extra watermark logits to the LLM's logits at each…

GumbelSoft: Diversified Language Model Watermarking via the GumbelMax-trick

2024-02-20 · Jiayi Fu, Xuandong Zhao, Ruihan Yang, Yuansen Zhang 외

Large language models (LLMs) excellently generate human-like text, but also raise concerns about misuse in fake news and academic dishonesty. Decoding-based watermark, particularly the GumbelMax-trick-based watermark(GM …

DiversityLanguage ModelingLanguage Modelling

On the Weaknesses of Backdoor-based Model Watermarking: An Information-theoretic Perspective

2024-09-10 · Aoting Hu, Yanzhi Chen, Renjie Xie, Adrian Weller

Safeguarding the intellectual property of machine learning models has emerged as a pressing concern in AI security. Model watermarking is a powerful technique for protecting ownership of machine learning models, yet its …

PostMark: A Robust Blackbox Watermark for Large Language Models

2024-06-20 · Yapei Chang, Kalpesh Krishna, Amir Houmansadr, John Wieting 외

The most effective techniques to detect LLM-generated text rely on inserting a detectable signature -- or watermark -- during the model's decoding process. Most existing watermarking methods require access to the underly…