Key facts
- Anthropic has implemented AI watermarking for its Claude chatbot.
- The watermarking embeds invisible signals into AI-generated text.
- This feature is intended to comply with the EU AI Act's transparency requirements.
- Some users have raised concerns about being potentially misidentified by the watermarking.
- The EU AI Act requires transparency regarding AI-generated content.
Anthropic has rolled out an AI watermarking feature for its Claude chatbot. This implementation is a direct response to the transparency requirements mandated by the EU AI Act. The watermarking technology embeds invisible signals within text generated by Claude. These signals are intended to help identify content produced by the AI. However, the introduction of this feature has not been without controversy. Some users have expressed concerns that the watermarking could lead to them being incorrectly identified as users of the AI tool, potentially causing issues for legitimate content creators. The EU AI Act's focus on transparency aims to ensure that users are aware when they are interacting with AI-generated content, and watermarking is one method to achieve this.