Anthropic watermarks Claude output under EU AI Act rules
What this answers
- Does Claude add a watermark to the text it generates?
- Yes — since August 2, 2026, Claude embeds invisible watermarks in generated text and signed provenance metadata in generated files, applied everywhere Claude is offered, not just the EU.
- Why is Anthropic watermarking Claude's output?
- To comply with the EU AI Act's Article 50(2) Code of Practice on Transparency of AI-Generated Content, which took effect August 2, 2026.
Anthropic began embedding invisible watermarks in Claude-generated text and signed provenance metadata in generated files on August 2, 2026, to comply with the EU AI Act’s Article 50(2) Code of Practice on Transparency of AI-Generated Content, per Chat GPT Is Eating the World. New Claude models launched from that date carry the marking by default, with earlier models being updated to add it; Anthropic said the marks apply wherever Claude is offered worldwide, not only in the EU.
Why it matters: As EU transparency rules phase in, Anthropic is applying AI-content marking globally rather than geo-fencing it to Europe, a compliance pattern other answer engines may face pressure to match.
The record: Anthropic
Via Chat GPT Is Eating the World ↗
Posted to the wire August 11, 2026. Edited by Joe Balewski.