[ Google Issues ] Study: How simple use of typos can disrupt LLM AI watermarks

0 views
Skip to first unread message

Lauren Weinstein

unread,
Feb 22, 2026, 12:03:05 PM (3 days ago) Feb 22
to google-is...@vortex.com

Study: How simple use of typos can disrupt LLM AI watermarks

https://www.ndss-symposium.org/ndss-paper/character-level-perturbations-disrupt-llm-watermarks/

- - -
--Lauren--
Lauren Weinstein
lau...@vortex.com (https://www.vortex.com/lauren)
Lauren's Blog: https://lauren.vortex.com
Mastodon: https://mastodon.laurenweinstein.org/@lauren
Signal: By request on need to know basis
Founder: Network Neutrality Squad: https://www.nnsquad.org
PRIVACY Forum: https://www.vortex.com/privacy-info
Co-Founder: People For Internet Responsibility
_______________________________________________
google-issues mailing list
https://lists.vortex.com/mailman/listinfo/google-issues
Reply all
Reply to author
Forward
0 new messages