Anthropic Adds Imperceptible Watermarks to Claude AI Text

Anthropic announced on Monday that its Claude AI models now embed imperceptible watermarks into generated text. The initiative aims to increase transparency and discourage the undetected use of artificial intelligence for academic work and novels.

The watermark remains intact during copying and pasting and can persist through some editing. However, the company noted that heavy paraphrasing, translation, or mixing the output with other text can render the watermark undetectable. This move aligns with the company's commitments under the European Union AI Act.

Models launched on or after August 2 already support the feature, and the company is working to integrate the technology into older models. Anthropic intends to provide third-party tools for watermark detection. This development follows a similar effort by Google DeepMind, which introduced SynthID for Gemini-generated content in 2024.