Anthropic watermarks all Claude outputs globally with marks that "may persist through some editing"

Anthropic is implementing invisible watermarks and C2PA digital signatures across all Claude-generated text and files to improve content transparency. This global initiative, which will be integrated into new models by August 2026, aims to help users identify synthetic media and comply with evolving European regulations. By providing third-party detection tools, the company intends to establish a standardized method for verifying the provenance of AI-created content. This shift reflects a broader industry move toward accountability as concerns grow regarding the potential for deepfakes and misinformation.