Anthropic expands Claude watermarking, but older models retain a loophole

Anthropic is adding text watermarking to four Claude models as it updates older systems ahead of the EU AI Act’s December deadline. Beginning November 16, 2026, Claude Opus 4.6, Claude Opus 4.5, Claude Sonnet 4.5, and Claude Haiku 4.5 are scheduled to receive the feature. The rollout on Amazon Bedrock may take several additional days. Claude Opus 4.6 had initially been expected to receive it on November 3.

The company says the watermark is an imperceptible pattern in word choice that does not add tokens, change pricing or formats, or include information about users or conversations. On Claude Opus 4.6 and Claude Sonnet 4.6, however, it is applied only when temperature, top_p, and top_k remain at their default settings. Changing any one of them can avoid the mark. Anthropic advises customers relying on watermarking for compliance to retain the defaults.

Claude Mythos 5 and Claude Opus 4.7 were listed in the earlier rollout and appear to have received the watermark on November 3. Anthropic uses a variant of Google DeepMind’s SynthID text method and has a detection API in private preview. The effect of watermarking on text quality remains disputed.