SynthID-Text watermark bypassed via invisible character injection
cyh-c · reddit · 2026-08-28
Tests show Google's SynthID-Text detects watermarks via token n-grams. By inserting default-ignorable characters (U+034F/U+FE00) after ASCII letters, detection drops from 188/192 to 0/192 while keeping visible text identical, revealing sensitivity to character-level tampering.
More from Safety
- CognitionX Dubai demo walks through full Microsoft Entra ID kill chain, from discovery to Global Admin — lee_stott · 2026-10-15
- DiPLab report: Scale AI's Outlier runs 16 trackers harvesting biometric data from EU workers — AntonioCasilli · 2026-10-15
- Your breathing gives you away: Wi-Fi signals can reveal your breath rhythm — TinfoilTricorn · 2026-10-15
- Cohere acquires Aisle, whose defensive cyber tech found 200+ High/Critical CVEs in curl, FFmpeg and more — aidangomez · 2026-10-15
- Podcast: AI is making pentesting more profitable and bug bounty more competitive — rez0__ · 2026-10-15
- Chatbots' bad health advice leaves no trace, warn researchers in Nature Health paper — alex_verem · 2026-10-15