Key facts
- Anthropic implemented AI watermarking for its Claude chatbot.
- The watermarking is intended to comply with the EU AI Act's transparency requirements.
- Invisible signals are embedded into AI-generated text by Anthropic's feature.
- Some users have expressed concern about being caught using the AI tool.
- Google will allow users to remove visible watermarks from AI-generated content.
- This applies to AI-generated images, videos, and songs from Google.
- Google will retain invisible watermarks and metadata for transparency.
- Google aims to balance creative control with safety.
Anthropic has rolled out a new AI watermarking feature for its Claude chatbot, a move designed to meet the transparency requirements of the EU AI Act. This new system embeds invisible signals within AI-generated text, allowing for the identification of content created by the chatbot. However, the implementation has faced some criticism from users who express concern about potentially being flagged for using the tool.
In a related development, Google has announced that it will allow users to remove visible watermarks from AI-generated content, including images, videos, and songs. The company clarified that while visible watermarks can be removed, invisible watermarks and associated metadata will remain in place. This approach aims to maintain transparency and accountability while also providing users with greater creative control over their AI-generated outputs. Google's strategy seeks to strike a balance between fostering creativity and ensuring safety and responsible AI deployment.