Abstract

Large Language Model (LLM) watermarking provides a lightweight mechanism for identifying text generated by a specific model, but its robustness remains fragile under postprocessing attacks. Deletion attacks are particularly challenging because they shift token positions and break the alignment between observed tokens and their original watermark positions. We propose **ReedMuller Code Watermarking (RMCW)** , an LLM watermarking method based on ReedMuller codes. In contrast to global codeword recovery, RMCW searches for surviving local algebraic structure, leveraging the ReedSolomon consistency induced by affine-line restrictions of Reed Muller codewords. During generation, RMCW injects a ReedMuller structure into the sequence via a secret-keyed vocabulary partition. During detection, it maps the given text to keyed vocabulary bins and tests local subsequences for low-degree ReedSolomon consistency using BerlekampWelch tests. Experiments on C4 and ELI5 datasets with OPT-1.3B and Llama-3.1-8B-Instruct show that RMCW preserves strong clean-text detectability and outperforms or matches the baseline methods under several deletion and rewriting attacks. Our code is available at https:// github.com/BaichengDanny/RMCW.