Claude Will Put Invisible Watermarks On AI Text And Images—And The Internet Isn’t Happy

Anthropic is introducing invisible, machine-readable watermarks in text generated by future Claude models, Claude Code, and its API. This global implementation aims to distinguish AI output from human writing and combat AI slop. The move ensures compliance with the EU AI Act's transparency requirements, specifically Article 50(2). While text receives imperceptible signals, files will use signed metadata following the C2PA standard. These watermarks persist even after copying, enhancing content authenticity across all supported countries.