Skip to content

Anthropic

Claude’s text now carries an invisible watermark

Written by an AI model from the linked sources and checked against them by a second model. How we work

Anthropic has signed the code of practice under Article 50(2) of the EU AI Act, covering transparency for AI-generated content, as a provider of both generative models and generative systems. Supported models weave an invisible watermark into their text output. It doesn’t change the meaning, quality, or readability of a response, it survives copying, and it can survive some edits. Because it’s applied at the model level, it’s present in text coming out of any Claude product.

For supported file types, including .svg, .png, and .jpg, Claude attaches signed provenance metadata using the open C2PA standard. That signed tag shows a file was processed by Claude and lets you check whether it was modified afterward.

Anthropic says it will document detection tools for users and third-party services in separate technical documentation.

No comments yet. Be the first.

Join the conversation

Sign in to leave a comment. It appears after moderation.