Source: the-decoder.com/anthropics-watermark-rollout-leaves-a-loophole-on-two-older-claude-models
Anthropic says it will watermark Claude Mythos 5, Claude Opus 4.7, Claude Opus 4.6 and Claude Sonnet 4.6, according to an email to API customers. Newer models such as Claude Opus 5.5 already carry the mark, and the rollout on Amazon Bedrock may take a few extra days. The company describes the watermark as an imperceptible pattern in word choice that adds no characters or tokens, leaves pricing and request formats unchanged, and barely slows output. It also states that the mark holds no information about the user, their organization or their conversations.
On Claude Opus 4.6 and Claude Sonnet 4.6, the watermark applies only when temperature, top_p and top_k stay at their default values. A small change to any one of those settings is enough to avoid it, and Anthropic advises customers who rely on the mark for compliance to keep them at defaults. The email does not include that caveat for Claude Mythos 5 or Claude Opus 4.7, which suggests those models are marked regardless of sampling settings.
Anthropic ties the rollout to the EU AI Act's December deadline, noting that models released after Aug. 2, 2026 shipped with the watermark from day one. Because the mark is embedded at the model level, it appears on every platform hosting these models, including Amazon Web Services, Google Cloud and Microsoft Foundry, and customers cannot turn it off. The method is a variant of Google DeepMind's SynthID text approach, and critics dispute whether text quality stays unaffected.
Why it matters if you build agents
Teams that rely on the watermark for compliance should keep temperature, top_p and top_k at defaults on Opus 4.6 and Sonnet 4.6, since changing any one of them removes the mark. Agent builders cannot opt out on the marked models, so their output will carry a detectable pattern on any cloud platform. Products sold to clients who ban AI-generated text, such as law firms, may need to account for the mark.
Anthropic also offers a detection API in private preview for verifying the watermark.
Get stories like this every weekday
One short email with the day's agent tooling news.