Anthropic recently made the decision to watermark Claude’s results, inserting invisible code into the chatbot’s editorial text that marks them as AI-generated.
Anthropic implemented this new policy to meet the EU AI Law Transparency Codewhich now requires technology companies to label AI-generated or edited content in a way identifiable to computer systems. However, while European regulators may be happy, some AI users are decidedly not.
One need look no further than Reddit to find evidence of the brewing discontent, although other posters on the site disagree. one of the most histrionic publications I came across a user named visionode, whose account is only three weeks old. According to visionode, the new watermarking system is a draconian conspiracy designed to victimize innocent chatbot users around the world.
Visionode’s basic argument seems to be that while savvy Claude users can hide evidence of their AI use by paraphrasing or scrubbing their results through other AI services, the average Claude user will be trapped.
“Who’s going to get caught? You. The student who used Claude to rearrange a paragraph. The journalist who asked the AI to summarize a two-hundred-page transcript. The writer who had creative block and asked for synonyms. Those guys come out of the process with a digital tattoo on their forehead.”
Far be it from me to try to undermine Visionode’s outrage, but those aren’t the best examples. A journalist asking AI to summarize a two-hundred-page transcript won’t mind a watermark attached to that summary, unless they are copying and pasting the summary word for word into their article, which is clearly unethical and shouldn’t be happening.
It is equally unethical for a student to copy and paste Claude’s production into an essay after asking him to “reorganize a paragraph.”
Other Redditors were also not particularly supportive of the poster’s outrage.
“Get a look at this guy,” one poster simply commented.
Another asked the OP to “take a deep breath.”
Visionode was not the only one to complain. Another disgruntled camper called the watermarks “unethical” and “disgusting” and argued that by using Claude, they had done most of the work. In his opinion, the chatbot was simply a “tool” that had made his hard work easier.
“I gave the instructions, the context, the decisions and the countless refinements, Claude was the tool. If Claude starts putting watermarks on the code or anything else he generates, why exactly is he claiming credit?” the poster asked.
Once again, other users attacked the reviewer.
“It’s not about claiming credit though,” one user responded. “It’s about being able to detect AI-generated results because of the risks that AI-generated results can cause in various situations.”
“Bro couldn’t even complain about Claude without using Claude to write it,” another joked.
Some critics have stayed away from the victim narrative and made slightly more nuanced arguments against Anthropic’s new policy.
For example, one poster complained of a general hypocrisy in putting a watermark on an editorial product that was itself generated by hoovering up other people’s work. “I think it is a very sinister direction” said the user. “I don’t use Claude to write anything, but having an AI watermark your work is terribly ironic given that many of the frontier models got their training data.”
However, in general, users have tended to support the watermarking system as a sensible way to track material generated by an algorithm. “There is literally no good argument why this is not a good idea,” said one user. in another thread he said. “The only reason you wouldn’t want this is to lie to people.”
When you purchase through links in our articles, we may earn a small commission. This does not affect our editorial independence.