SWAN: Semantic Watermarking with Abstract Meaning Representation

TL;DR

SWAN introduces a semantic watermarking framework embedding signatures into sentence structures via Abstract Meaning Representation, enhancing robustness against paraphrasing and enabling prompt-based, training-free text provenance verification.

Contribution

It is the first to embed watermarks directly in semantic structures using AMR, improving robustness and simplicity over existing token-based methods.

Findings

SWAN achieves state-of-the-art detection performance on unaltered text.

It significantly improves robustness against paraphrasing, with up to 13.9% higher detection AUC.

The method is training-free and relies on prompt-guided generation and AMR parsing.

Abstract

We introduce SWAN (Semantic Watermarking with Abstract Meaning Representation), a novel framework that embeds watermark signatures into the semantic structure of a sentence using Abstract Meaning Representation (AMR). In contrast to existing watermarking methods, which typically encode signatures by adjusting token selection preferences during text generation, SWAN embeds the signature directly in the sentence's semantic representation. As the signature is encoded at the semantic structure level, any paraphrase that preserves meaning automatically preserves the signature. SWAN is training-free: watermark injection is achieved by prompting an LLM to generate sentences guided by a selected AMR template while maintaining contextual coherence, and detection uses an off-the-shelf AMR parser followed by a simple one-proportion z-test. Empirical evaluation on the RealNews benchmark shows SWAN matches state-of-the-art detection performance on unaltered watermarked text, while significantly improving robustness against paraphrasing, increasing detection AUC by up to 13.9 percentage points compared to prior methods. These results demonstrate that SWAN's approach of anchoring watermarks in AMR semantic structures provides a simple, effective, and prompt-based method for robust text provenance verification under paraphrasing, opening new avenues for semantic-level watermarking research.