Anthropic says it will add text watermarks to AI-generated output
Anthropic says it plans to watermark text generated by its AI models and extend the feature to output from older models.
In short
- Anthropic says it plans to watermark text generated by its AI models and extend the feature to output from older models.
Anthropic says it will introduce watermarking for text generated by its artificial-intelligence models, according to a TechCrunch report published on August 11, 2026.
The company also plans to extend watermarking support to generations produced by older models, the report said. The available source record does not specify a rollout date, the technical method used to mark text or which older models will be covered.
Text watermarking is intended to provide a signal that content may have been produced by an AI system. The effectiveness of any implementation will depend on details such as how reliably the signal survives editing and how it can be detected; Anthropic’s reported announcement, as captured in the source record, does not address those questions.
