Anthropic has begun inserting invisible watermarks into texts generated by the Claude chatbot, in a move to comply with the Transparency Code of the European Union's AI Act. The decision, revealed by TechCrunch, drew negative reactions from some users, with criticism posted on Reddit.
The new policy requires technology companies to identify, in a machine-readable way, content produced or edited by artificial intelligence. The watermark is an invisible code embedded in the text.
Reactions on Reddit
Among the critics, a user identified as visionode called the initiative a "draconian conspiracy" against ordinary users. In his view, students who use the tool to reorganize paragraphs, journalists who summarize long transcripts, and writers looking for synonyms would be the main targets, ending up with a "digital tattoo on their foreheads."
visionode's post did not receive widespread support. Other Reddit members responded with mockery, including requests for the author to "take a deep breath." One of them commented that the complainant couldn't criticize Claude without using Claude itself to write the complaint.
Another critic called the watermarks "unethical" and "disgusting," arguing that the user provides instructions, context, and decisions, leaving the chatbot to merely execute. A defender of the measure countered that the goal is not to give credit to AI, but to allow detection of algorithm-generated content, given the risks they may pose.
A third critic pointed to an alleged contradiction in the policy, citing the fact that AI models were trained on other people's data. "I think it's a very sinister direction to take," they said, calling the watermark "terrifyingly ironic."
However, most comments were favorable to the measure. "There is literally no good argument against it," said one user. "The only reason not to want it is to lie to people."
