Anthropic announced it will add machine-readable watermarks to content processed by its AI models, not just generated outputs, to comply with the EU’s AI Act. All new models released globally will include embedded watermarks from day one, with generated files carrying digitally signed provenance metadata where supported.
Although the EU does not require watermarking for minor assistive edits like grammar correction, Anthropic is applying it more broadly. These markers will be invisible to users, at least for the time being.
Comments