Anthropic will add watermarks to content generated by models such as Claude to comply with EU AI Act requirements.

TechFlow News, August 11, according to TechCrunch, Anthropic stated that it will add identifiable watermarks to text and files generated by its models to comply with the transparency regulations of the effective EU "AI Act". The company stated that all models released after August 2 will have this mechanism enabled by default, with file watermarks adopting the C2PA open standard, text watermarks remaining intact after copy-pasting, and the feature will be gradually extended to older models. This measure will cover products such as Claude Platform API, Claude, and Claude Code.