Anthropic's text watermarks signal new front in AI detection
TL;DR
Anthropic's new models will add machine-readable "watermarks" to Claude-generated text and files to comply with new European Union transparency regulations. Why it matters: Comms teams using Claude to simply clean up, translate, or format human-drafted press releases could stamp those documents with an AI signature. How it works: For models launched in the EU after Aug. 2, Anthropic is marking content in two ways, "wherever Claude is offered, worldwide.
Nauti's Take
Teams using Claude for press releases or client material should test their own documents before rollout to see when a machine-readable mark appears and which metadata files carry. Sensitive communications need a clear approval rule covering allowed edits, disclosure, and whether recipients can inspect the signature.
Anthropic’s detection claims and the exact implementation still need independent verification.
Sources