The EU AI Act transparency rules took effect on August 2, requiring all new models to embed machine-readable identifiers.

Anthropic will watermark text and files generated by Claude and other AI models to comply with the EU AI Act. The Transparency Code took effect on August 2 and requires machine-readable labels for AI-generated content.

Anthropic confirmed the plan in a support page update, joining Google, Meta, Microsoft, OpenAI, Synthesia, and Black Forest Labs in committing to the EU code. The company will use the C2PA open standard for files and extend the capability to older models.

Developers integrating Claude must expect watermarks in every generated response, whether from the API, Claude Code, or other surfaces. Anthropic applies the watermark at the model level, so it persists through copy and paste and may survive light editing. Teams should factor this into content pipelines and moderation workflows.

Anthropic has not specified how much editing removes the watermark. The company says it will extend the feature to older models, but has not given a timeline. Other platforms, including Suno and Substack, have also introduced content labeling in response to regulatory and user pressure.

What matters

• Anthropic will add watermarks to text and files from all Claude models released after August 2.
• Builders using the Claude API must now plan for watermarks in every generated response.
• Anthropic has not said how much editing removes the watermark, but older models will eventually gain support.

Why it matters

Anthropic has not said how much editing removes the watermark, but older models will eventually gain support.

This GenAI News article was prepared in original wording using reporting and materials published by TechCrunch AI. Source reference: https://techcrunch.com/2026/08/11/anthropic-says-it-will-watermark-text-generated-by-its-ai-models/.

Drafted by the GenAI News review pipeline.