Invisible watermarks on AI text: Why Anthropic’s latest move has got writers worried | AI generated

Anthropic’s latest Claude models are embedding invisible, machine-readable watermarks into generated text—marks that can survive copy-pasting and some editing. But the watermark doesn’t necessarily prove that a piece was written entirely by AI. It can also appear when Claude is used to edit, translate or summarise human-written content. So, what does this mean for writers and AI users? And why is Anthropic introducing these markings now? This explainer breaks down AI provenance, invisible watermarks and the growing debate over AI-assisted writing.