EU AI Act Pressure: Anthropic Embeds Watermarks into Claude Outputs Globally

neil_chilson · x · 2026-08-12

To comply with Article 50 of the EU AI Act, Anthropic announced it is embedding an invisible watermark directly into Claude-generated text. Because the mark is woven into the words themselves, it persists through copy-pasting and survives later editing. Anthropic plans to apply this across Claude, Claude Code, its API, and cloud platforms globally. Commentators note this as a clear example of the Brussels Effect influencing AI model outputs worldwide, one token at a time.

Related event: Anthropic Introduces Invisible Watermarks for Claude to Comply with EU Regulations(77 posts)→

Original post →

More from Models

Models channel →