Anthropic Puts Hidden Watermarks on Claude Text Under New EU Rules
Anthropic embeds invisible watermarks in Claude text worldwide to meet EU transparency rules, with marks designed to survive copy-paste and some edits.
- Anthropic implemented machine-readable watermarks for New Claude models to tag AI-generated content worldwide, ensuring transparency across products like Claude Code and Claude Cowork.
- The initiative aligns with the European Union Code of Practice and transparency requirements under the Act, which Anthropic signed to standardize AI-generated content tracking.
- Anthropic utilizes invisible signals for Text, while Files carry signed provenance metadata based on the C2PA standard to record Content Provenance and Authenticity.
- Platforms including AWS, Google Cloud, and Microsoft Foundry now incorporate these markings, providing a machine-readable identity that helps third parties detect Claude-generated material.
- Detection reliability remains limited for both Text and Files, as Anthropic cautions that editing or Short passages can compromise the signals and undermine definitive authorship proof.
31 Articles
31 Articles
"The meaning, quality and legibility of the response is not seen and does not change," says the company, which has not yet explained how it will be possible to verify whether a particular text carries that 'generated with AI' signal.
The vast majority of Canadian authors will not receive compensation from Anthropic, who used their books to train Claude.
Claude AI Watermark: Anthropic Adds Invisible Markers to AI-Generated Text and Files, Know How It Works | 📲 LatestLY
Anthropic has integrated invisible watermarks and metadata into Claude models to comply with the EU AI Act. According to an Anthropic post, new models will feature machine-readable markers from day one, allowing third-party tools to detect AI-generated text and media while preserving normal reading quality. 📲 Claude AI Watermark: Anthropic Adds Invisible Markers to AI-Generated Text and Files, Know How It Works.
Anthropic rolled out a feature that stops undetected AI-generated writing
Anthropic is introducing a new feature to improve transparency on AI-generated content.Samuel Boivin/NurPhoto via Getty ImagesAnthropic will now embed invisible watermarks in Claude AI text.It is part of Anthropic's commitments under the EU AI Act but applies to text generated worldwide.Anthropic said that heavy editing can erase the mark, and a watermark does not guarantee AI use.It's going to get a lot harder to let Claude be a ghostwriter for…
Coverage Details
Bias Distribution
- 40% of the sources lean Left
Factuality
To view factuality data please Upgrade to Premium
