Anthropic’s chatbot, Claude, now embeds an imperceptible watermark into the text it generates. This development aims to combat the increasing misuse of AI-written content, like plagiarism and the spread of misinformation. The watermark allows for identification of text originating from Claude, even after copying and pasting. This means users who carelessly reproduce AI-generated text risk being detected. The technology functions by subtly altering word choices during text creation, creating a statistical signature. While not foolproof, the watermark presents a significant step toward increasing transparency regarding AI-generated content and its origins. Developers hope this feature will mitigate the potential harms associated with undetectable AI writing.