Abstract
Semantic-level watermarking (SWM) enhances robustness against text modifications by treating sentences as the basic unit. However, robustness to paragraph-level paraphrasing remains challenging as such attacks globally disrupt watermark signals by changing sentence order. In this work, we propose SAMark, a self-anchored watermarking framework that removes the dependency on sentence order by establishing an order-independent green region in semantic space.
To improve detectability, we introduce a multi-channel hyperbolic scoring mechanism that amplifies watermark signals while suppressing noise from weakly aligned candidates. We further propose a diversity-aware filtering strategy that combines hard filtering with soft regularization, extending beyond simple n-gram repetition filters to address semantic redundancy.
Experimental results show that SAMark achieves up to 90.2% TP@FP1% under typical paragraph-level paraphrasing attacks, outperforming the strongest prior baseline by more than 30% on average, while maintaining generation quality competitive with unwatermarked text and breaking the robustness-quality trade-off that limits prior methods. Our code will be released at this URL.
Blogger's Review: SAMark's self-anchored design showcases significant resistance against paragraph-level paraphrasing attacks, marking a notable advancement in watermark robustness. This framework not only enhances the effectiveness of text watermarking but also provides new insights for future text protection technologies. The multi-channel hyperbolic scoring and diversity filtering effectively reduce semantic redundancy while maintaining generation quality, making it a noteworthy development.