Anthropic has introduced a watermark to text produced by its AI, Claude. This move is in compliance with the new EU AI Act. The watermark is integrated into newer models of Claude, with implementation for older models underway.

The watermark aids in detecting text generated by Claude, but is not noticeable to human readers. An AI researcher describes the watermark as "brilliant" and notes that it is complex and cannot be easily removed. Further details about the watermark and its implementation are available from the source.

The introduction of this watermark is a significant development in AI technology, particularly in terms of transparency and detection.