Anthropic introduces hidden watermarks in Claude’s AI text output

Anthropic has begun embedding invisible, machine-readable watermarks into text generated by Claude, creating a signal that can travel with the content even after copying. According to Nairametrics, the company said the move supports its commitments under the European Union’s AI Act Code of Practice on Transparency of AI-Generated Content.

Claude models launched on or after August 2, 2026, support the feature from launch, while Anthropic works to extend it to earlier models. The company noted the watermark may not survive heavy editing, paraphrasing or translation, and its presence or absence is not conclusive proof of authorship.