How Anthropic plans to watermark Claude's AI-generated text

Anthropic is rolling out invisible watermarking for text generated by Claude, making it one of the first AI companies to detail how it will meet the EU's requirement to label AI-generated content. The watermark works by subtly influencing word choices during text generation using a secret key, leaving a detectable statistical pattern without adding hidden characters or slowing output. A detection API is planned, though it can only estimate Claude's involvement, not prove authorship.