Anthropic to Embed Invisible Watermarks in Generated Text to Comply with EU AI Act
EricBuess · x · 2026-08-12
Anthropic has announced that, to comply with the EU AI Act codes, its new Claude models will embed invisible watermarks in all generated text across regions where Claude is available.
- Watermark Features: The watermark is part of the text itself, not just metadata. This means it will travel with the text when copied and pasted elsewhere and may persist through some editing.
- Rollout Plan: The requirement applies to models launched on or after August 2 and will be rolled out worldwide. Anthropic is currently working on adding this feature to existing models and plans to ship a text detection API for public use.
Related event: Anthropic Adds Invisible Watermarks to Claude Outputs for EU Compliance(3 posts)→
More from Models
- Musk Jokes About US Debt, Grok Chimes in with $18 Trillion Estimate — Kuprel · 2026-08-12
- Google Gemma Hits 1 Billion Downloads Milestone — lateinteraction · 2026-08-12
- Independent AI Labs Surge: Token Share Exceeds Top Incumbents Combined — shensi · 2026-08-12
- SGLang Enables Local Deployment of Nemotron 3.5 with 1M Context — BanghuaZ · 2026-08-12
- GPT and Claude Settle a 25-Year-Old Information Theory Problem — weijie444 · 2026-08-12
- Enterprise AI Shift to Specialized Small Models: Generic LLMs Waste 99% of Compute — blaizedsouza · 2026-08-12