{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,12,23]],"date-time":"2025-12-23T10:05:06Z","timestamp":1766484306243,"version":"3.41.2"},"reference-count":0,"publisher":"Privacy Enhancing Technologies Symposium Advisory Board","issue":"4","license":[{"start":{"date-parts":[[2025,10,1]],"date-time":"2025-10-01T00:00:00Z","timestamp":1759276800000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["PoPETs"],"abstract":"<jats:p>Large language models (LLMs) have demonstrated remarkable performance in various tasks, but they also face challenges in intellectual property (IP) protection. Traditional training-based watermarking techniques are computationally expensive, while function invariant transformations (FITs) offer a lightweight alternative. Nevertheless, FIT-based watermarking methods are vulnerable to adaptive attacks, where adversaries can exploit the same transformation to remove or forge watermarks. We propose a novel white-box watermarking scheme that combines error correction codes (ECCs) with weight permutations. By encoding model identifiers using ECCs, our approach guarantees reliable watermark extraction under various attacks. Additionally, we develop a linear assignment-based extraction algorithm to enhance its efficiency. Evaluations on six LLMs show that our method offers robust watermarking capabilities. It has a minimal impact on model performance while effectively defending against removal and forgery attacks. Overall, our approach provides a scalable and secure solution for safeguarding the copyrights of LLMs.<\/jats:p>","DOI":"10.56553\/popets-2025-0126","type":"journal-article","created":{"date-parts":[[2025,7,13]],"date-time":"2025-07-13T15:58:27Z","timestamp":1752422307000},"page":"183-200","source":"Crossref","is-referenced-by-count":2,"title":["Robust and Efficient Watermarking of Large Language Models Using Error Correction Codes"],"prefix":"10.56553","volume":"2025","author":[{"given":"Xiaokun","family":"Luan","sequence":"first","affiliation":[{"name":"Peking University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zeming","family":"Wei","sequence":"additional","affiliation":[{"name":"Peking University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yihao","family":"Zhang","sequence":"additional","affiliation":[{"name":"Peking University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Meng","family":"Sun","sequence":"additional","affiliation":[{"name":"Peking University"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"35752","published-online":{"date-parts":[[2025,10]]},"container-title":["Proceedings on Privacy Enhancing Technologies"],"original-title":[],"deposited":{"date-parts":[[2025,7,13]],"date-time":"2025-07-13T15:58:29Z","timestamp":1752422309000},"score":1,"resource":{"primary":{"URL":"https:\/\/petsymposium.org\/popets\/2025\/popets-2025-0126.php"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10]]},"references-count":0,"journal-issue":{"issue":"4","published-online":{"date-parts":[[2025,10]]}},"alternative-id":["10.56553\/popets-2025-0126"],"URL":"https:\/\/doi.org\/10.56553\/popets-2025-0126","relation":{},"ISSN":["2299-0984"],"issn-type":[{"type":"electronic","value":"2299-0984"}],"subject":[],"published":{"date-parts":[[2025,10]]}}}