Anthropic to watermark text generated by Claude and other AI models to meet EU rules

Anthropic will watermark text generated by its AI models, including Claude, to comply with European regulations, the company confirmed in an updated support page. The watermarking technology, which applies to all models released after August 2, is designed to meet the EU AI Act’s Transparency Code, which requires AI companies to mark AI-generated or edited content in a way other systems can identify.

How Anthropic’s watermarking works

Anthropic said that all models released after August 2 will automatically include technology that watermarks both computer-generated text and files. For files, the company is using the C2PA open standard. The watermark is embedded at the model level, meaning it will be present regardless of which Claude product or surface the text comes from, including the Claude platform API, Claude, Claude Code, Claude Cowork, and Claude Tag.

According to the support page, the watermark travels with the text when it’s copied and pasted elsewhere, and may persist through some editing. However, the company did not specify how much editing might remove the watermark. Bitcoin World has asked Anthropic for clarification and will update this story if more details are provided.

Why this matters for AI regulation and users

The move is part of a broader industry trend. Platforms are rushing to watermark AI-generated content after user backlash and to avoid regulatory scrutiny. Last week, AI music platform Suno said it will mark tracks created on its platform after legal challenges. Last month, Substack teamed up with Pangram to flag AI-generated content, with CEO Chris Best calling out “Claudefishing,” a term for using AI to generate content.

Anthropic joins other companies like Black Forest Labs, Google, Meta, Microsoft, OpenAI, and Synthesia in committing to the EU’s code. The EU AI Act’s transparency requirements took effect on August 2, and this watermarking is a direct response to that legal framework.

What this means for content creators and consumers

For everyday users, watermarking adds a layer of transparency, helping identify AI-generated text. For businesses and creators who rely on AI tools, it could affect how they use and share content, especially in contexts where authenticity matters. The long-term impact on AI adoption and content verification remains to be seen, but this is a significant step toward standardized labeling.

Conclusion

Anthropic’s decision to watermark text from its AI models is a proactive move to align with EU regulations and address growing concerns about AI content transparency. As more companies adopt similar measures, the industry is moving toward a more accountable ecosystem. The effectiveness of these watermarks in real-world scenarios will depend on how well they withstand editing and how widely they are adopted across platforms.

FAQs

Q1: What is the EU AI Act’s Transparency Code?
The Transparency Code is a part of the EU AI Act that requires AI companies to mark AI-generated or edited content in a way that other systems can identify. It took effect on August 2.

Q2: How does Anthropic’s watermarking work?
Anthropic embeds a watermark at the model level, so any text generated by its models, including Claude, carries the mark. The watermark travels with the text when copied and may persist through some editing.

Q3: Will older Anthropic models be watermarked?
Anthropic said it will extend support for older models as well, though details on timing and implementation are not yet fully clear.