Anthropic has begun inserting hard-to-detect watermarks into outputs from some Claude models, raising concerns over copyright and privacy. The watermark can remain after copy-and-paste and may persist after edits, and Anthropic says it is meant to distinguish human and AI writing. Critics warn the company could act as sole arbiter, while developers question impacts on text quality and on lightly edited human writing. Anthropic says it will provide a free API and that the mark indicates processing, not authorship.

Anthropic has introduced a feature that embeds a watermark that is difficult to detect visually into text written by its generative AI model Claude. The move aims to help identify cases where AI-generated writing is submitted or published as if written by a person, and could affect how the education and publishing industries verify AI use. The watermark does not change meaning or readability and is designed to persist through copying and some edits.