How Claude marks AI-generated content

	How Claude marks AI-generated content(support.claude.com)
	443 points by mfiguiere 11 days ago \| 408 comments
	tl;dr: Anthropic will embed imperceptible watermarks into Claude-generated text and attach C2PA-signed provenance metadata to generated files (SVG, PNG, JPG), applied at the model level so marks persist across Claude products. The company is also developing detection tools for third parties to verify whether content was produced by Claude. This implements Anthropic's commitments under the EU AI Act's Article 50(2) Code of Practice, though developers building on Claude must independently assess their own transparency obligations.
	HN Discussion: ~Concern about false positives leading to unfair accusations against human writers ↓Watermarking will degrade output quality by biasing token selection away from optimal choices ↓Text watermarking is fundamentally unreliable and shouldn't be trusted for detection ↓Hybrid human-AI workflows will be unfairly flagged, making Claude unusable for legitimate use cases •Curiosity about technical implementation and competitive market dynamics of watermarking