Anthropic has announced that text created by its AI models, including Claude, will now be watermarked to align with European transparency rules. The company confirmed the change through an updated support page as the EU AI Act's transparency requirements take effect.
According to Anthropic, models released after August 2 will automatically include technology that marks both generated text and files. For documents, the company says it is using the C2PA open standard, a framework designed to help identify digital content provenance.
The watermark is embedded at the model level, which means it can remain attached when text is copied and pasted into other places. Anthropic also says older models will receive support, and the system will apply across products such as the Claude platform API, Claude, Claude Code, Claude Cowork, and Claude Tag.
The move reflects a wider shift across the AI industry as companies adapt to transparency expectations and content authenticity standards. Other major technology firms have also signaled support for the EU's approach, reinforcing a broader push toward clearer labeling of AI-generated material.
As AI tools become more deeply woven into everyday work, watermarking may help build a more traceable and trustworthy digital future.
