Anthropic models will soon inject watermarks identifying AI-generated text

# Anthropic's AI Will Soon Mark All Its Text as Human-Made Starting next year, Anthropic's Claude AI will add an invisible watermark to everything it writes so readers can tell it came from AI rather than a person—and this watermark will stick with the text even if someone copies and pastes it elsewhere. The company is doing this to comply with new European privacy laws, though it plans to apply the same rule everywhere Claude is used. This means if you use Claude for work, people reading your documents will be able to see that parts of it came from AI.