Some users of the Claude chatbot have not appreciated the appearance of watermarks on generated content. Some fear being caught using AI, while others view the watermarking as an infringement on copyright, reports TechCrunch.
Since August 2, 2026, Anthropic has been implementing watermarking for content created by Claude. An invisible watermark is embedded into the generated text, which does not affect meaning or readability but persists even when copied to other platforms.
One of the most emotional posts on this matter was published by Reddit user visionode. According to him, the new watermarking system is a 'conspiracy designed to harm chatbot users worldwide.' He notes that while experienced users will be able to hide traces of AI use, the average user will be caught.
'Who will get caught? You. The student who used Claude to reorganize a paragraph; the journalist who asked AI to summarize a 200-page transcript; the writer who had a creative block and asked for synonyms. These folks will end up with a digital tattoo on their foreheads.'
Another dissatisfied user called the watermarks 'unethical' and 'disgusting.' In his opinion, they imply that Claude did the lion's share of the work.
'I provided prompts, context, solutions, and countless revisions; Claude was a tool. When Claude puts watermarks on code or other generated content, what authorship is it claiming?' asked the post's author.
Another user believes that putting watermarks on a product that was itself created by borrowing others' work is hypocrisy.
'I don't use Claude, but the very fact that AI puts watermarks on your work is so ironic, given how many advanced models were trained.'
Earlier, a Japanese developer introduced the Claude Watermark Remover resource to remove crypto-tags from Claude.
