Anthropic debunks Claude watermarking myths: SynthID can't fingerprint or track users
DevDminGod · x · 2026-09-06
Guillaume Meyer published an article debunking top myths about Claude watermarking (SynthID), starting with the fact that the technology carries only an extremely limited number of bits — not enough to fingerprint or identify individual users. Commenter DevDminGod questions why Anthropic keeps landing itself in watermarking-related controversies when only a handful of top labs are training models.
More from Safety
- Boaz Barak on Astra alignment gains: gap with capabilities still widening, warns Greenblatt — dhadfieldmenell · 2026-09-06
- Brundage: policymakers will eventually freak out that AI labs can't defend their IP — Miles_Brundage · 2026-09-06
- Survey: 78% of national security experts see AI boosting cyber defense, humans the bottleneck — elie · 2026-09-06
- Survey: 78% believe AI will ultimately improve cyber defense capabilities — elie · 2026-09-06
- Frontier AI models fix only 1 in 4 security vulnerabilities correctly, report finds — Evgenii42 · 2026-09-06
- Jack Clark praises GDM's agent-swarm emergent properties paper, teases Import AI writeup — jackclarkSF · 2026-09-06