Anthropic has started adding machine-readable watermarks to Claude-generated content to comply with the EU AI Act’s rules, with the watermarks eventually rolled out worldwide. Some AI power users who secretly use chatbots for work or study are upset, arguing the labels could stigmatize people who use AI for writing, studying, or work. Supporters say watermarks help identify AI-generated output and reduce misuse in schools, hiring, journalism, and research. A Reddit user named visionode claimed the move is a conspiracy to punish poor serial prompters, adding that while smart users will find a way to avoid getting caught, most won’t, and that it creates a “caste of ‘dirty’ creators.” Anthropic has acknowledged that if a watermark is detected, it doesn’t necessarily mean the entire text, image, or file is AI-generated—it may have just been processed by Claude. The article notes that using AI to generate a summary of a transcript is different from copying and pasting the chatbot’s output into an article, which is plainly unethical, as is using Claude’s content in an essay. It also cites that AI-powered cheating is off the scale, mentioning that last month, a Brown University professor saw his students’ midterm scores significantly drop when he changed the exam to an in-person test where AI couldn’t be used.