Signals · Hot · first seen October 30, 2026
Anthropic is adding its hidden text watermark to four older Claude models
In an email to API customers, Anthropic said it will watermark text from Claude Mythos 5, Opus 4.7, Opus 4.6 and Sonnet 4.6, with Amazon Bedrock a few days behind, The Decoder reported on Oct. 31. Models released since Aug. 2 already carry the mark, and Anthropic points to the EU AI Act's December deadline. On Opus 4.6 and Sonnet 4.6, it applies only when the sampling settings are left at their defaults, and customers cannot turn it off.
Why it matters to a business like yours
A customer email, a job posting or a website page drafted with Claude can now carry a pattern that a checking tool may pick up. Decide whether you tell customers when AI helped write something, and put that answer in your staff AI rules before a customer asks.
Sources
- Anthropic's watermark rollout leaves a loophole on two older Claude models (The Decoder, October 31, 2026)
On the Insider
- Claude Marks the Text It Writes. Decide What You Tell Clients. (November 1, 2026)
History
- November 2, 2026: Hot