Anthropic announced that it will embed machine-readable watermarks in all content processed by its Claude models, not just generated outputs. The move aligns with the EU AI Act, which requires AI-generated or manipulated content to be marked, with a grace period for older models until December 2026.
The watermark will be invisible to users, while generated files will get digital provenance metadata where supported. Anthropic is applying the watermark globally and to all processed content, going beyond the EU’s exemptions for standard editing or minor alterations.
Comments