May 3, 2024, 4:16 a.m. | John Kirchenbauer, Jonas Geiping, Yuxin Wen, Jonathan Katz, Ian Miers, Tom Goldstein

cs.CR updates on arXiv.org arxiv.org

arXiv:2301.10226v4 Announce Type: replace-cross
Abstract: Potential harms of large language models can be mitigated by watermarking model output, i.e., embedding signals into generated text that are invisible to humans but algorithmically detectable from a short span of tokens. We propose a watermarking framework for proprietary language models. The watermark can be embedded with negligible impact on text quality, and can be detected using an efficient open-source algorithm without access to the language model API or parameters. The watermark works by …

arxiv can cs.cl cs.cr cs.lg embedded framework generated humans language language models large signals span text tokens watermarking

Information Security Engineers

@ D. E. Shaw Research | New York City

Technology Security Analyst

@ Halton Region | Oakville, Ontario, Canada

Senior Cyber Security Analyst

@ Valley Water | San Jose, CA

Security Operations Manager-West Coast

@ The Walt Disney Company | USA - CA - 2500 Broadway Street

Vulnerability Analyst - Remote (WFH)

@ Cognitive Medical Systems | Phoenix, AZ, US | Oak Ridge, TN, US | Austin, TX, US | Oregon, US | Austin, TX, US

Senior Mainframe Security Administrator

@ Danske Bank | Copenhagen V, Denmark