Anthropic is reportedly adding hidden watermarks to text generated by Claude AI, making it easier to identify AI-created content.
The markers are embedded in generated text without visibly changing how the content appears to users.
The technology could allow Anthropic or authorized tools to determine whether a piece of text was produced by Claude.
The move comes as AI-generated content becomes increasingly common across websites, education, publishing, and social media.
AI text detection can be challenging, particularly when generated content is edited, translated, or rewritten by humans.
Hidden watermarking could become an important tool for identifying AI-generated material while improving transparency around synthetic content.