Anthropic is now embedding invisible watermarks into all text generated by Claude and attaching signed provenance metadata to all files it creates.

The watermarks are woven directly into the text, they persist through copy/paste and even some editing. For files like images and SVGs, Claude attaches C2PA metadata that signals the content was AI-generated and flags any tampering.

This is being done to comply with the EU AI Act's transparency requirements, but the implications go beyond Europe.

Every piece of text Claude produces now carries a hidden fingerprint that can be detected by anyone with the right tools.

We're entering an era where AI-generated content is marked at the source whether you want it to be or not.