Lost in Overlap: Exploring Watermark Collision in LLMs

arXiv:2403.10020v1 Announce Type: new
Abstract: 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 use of watermarking across diverse LLMs has led to an inevitable issue known as watermark collision during common tasks like question answering and paraphrasing. This study focuses on dual watermark collisions, where two watermarks are present simultaneously in the same text. The research …

abstract arxiv challenges collision concerns copyright cs.cl cs.mm diverse embed however issue language language models large language large language models llms lost raises text type watermark watermarking