Anthropic's Invisible 'Scarlet Letter' Watermark for Claude Is Coming

Anthropic has announced that it will embed invisible, machine-readable watermarks into content processed or generated by its AI models, starting with all new models offered globally. The move goes beyond the EU AI Act's requirements, which already exempts some assistive functions like grammar correction.

Text outputs will carry hidden watermarking, while supported files will include digitally signed provenance metadata. Pre-existing models have until December 2026 to comply.

Get the app vidgetc Tech, gaming & AI news — always at hand Google Play · Soon App Store
Discuss

Comments

Next articleLe Figaro reporting is coming to ChatGPT→
Start typing to search