Claude's invisible text watermark uses SynthID, embedded token by token—and already beatable
APPSO · wechat · 2026-08-17
Anthropic confirmed Claude's text watermarking uses Google DeepMind's SynthID. Rather than stamping text after generation, it subtly adjusts candidate token scores at every generation step, creating a detectable statistical pattern only visible across many tokens—like a teacher hinting "when in doubt, pick C". Meanwhile, removal tools like Declaude have already appeared on GitHub: paraphrasing via another model can break the statistical signal. The piece also raises a deeper concern voiced by John Gruber: watermarking adds an optimization goal unrelated to writing well, and the arms race over rewording squeezes out care for style and precision in language itself.
More from Models
- Meta releases Muse Code; Alibaba open-sources Qwen3.8-27B — emmanuelvivier · 2026-08-17
- Meta open-sources 30B-parameter agent model Muse Glimmer under Apache 2.0, runs on consumer GPUs — emmanuelvivier · 2026-08-17
- Netizen jokes Qwen 3.8 27B defaults to 'overthinking' — emax · 2026-08-17
- Claude's coding 'one-shots' often just copy-pasted from humans — art_zucker · 2026-08-17
- Fable Coding Test: Elegant Solution Outperforming Opus 5 — arjunrajlab · 2026-08-17
- DeepSeek's price hike lands Sunday: V4-Flash peak output jumps ~4.7x — NeuralNomad87 · 2026-08-17