Anthropic adds text watermarking to latest Claude models
In short: Anthropic has implemented a text-based watermark for its newest Claude models, including Opus 4.6, Opus 4.5, Sonnet 4.5, and Haiku 4.5. The watermark applies when using default sampling configurations but may be absent if users modify temperature, top_p, or top_k settings.
This summary was generated automatically by AI from Claude Platform Release Notes's publication. It is our own text, not a copy of the original — facts, figures and quotes belong to the source, linked above and below.
What changed?
- 1Introduction of watermark for Claude 4.5 and 4.6 families
- 2Watermark exclusion if temperature, top_p, or top_k deviate from defaults
Why it matters
This affects organizations implementing AI-generated content policies or provenance tracking. Developers must note that modifying sampling parameters effectively removes the model's integrated provenance marker.
What it means for AI agents and contact centers
For teams building voice agents, ensure you monitor how watermarking impacts output generation. If you rely on automated content verification, be aware that tuning creativity parameters for voice synthesis or dialogue may disable the native watermark mechanism.
🧪 Worth Testing
Verify whether your production API calls use default sampling parameters and confirm how the watermark manifests in your logs or output processing pipelines.
Sources
- Claude Platform Release NotesOfficialPrimary sourceOriginal article →„Text generated by Claude Opus 4.6, Claude Opus 4.5, Claude Sonnet 4.5, and Claude Haiku 4.5 carries Anthropic’s text watermark on supported platforms. How Claude marks AI-generated content lists those“16 Nov 2026, 02:00
- Published by source
- 16 Nov 2026, 02:00
- Found by our system
- 16 Nov 2026, 22:08
- Summary generated
- 16 Nov 2026, 22:08
This article was written by AI from the original source. Facts, numbers and prices come from the source; missing values are marked “Not specified”. Legal notice, copyright and privacy