Claude has started to incorporate invisible watermarks in its texts to identify content generated by artificial intelligence.
The measure by Anthropic aims to comply with the new transparency requirements of the European Union and has generated criticism among some users.
Claude began to incorporate invisible watermarks into his texts
Why Anthropic incorporated watermarks in Claude
Anthropic decided to incorporate an invisible watermark in the content generated by Claude. This involves code inserted in the texts that allows them to be identified as material generated by AI.
The company implemented this measure to comply with the EU AI Act and its transparency code. The regulation requires that content generated or edited using artificial intelligence can be identified by computer systems.
What criticisms the new Claude system received
On Reddit, various criticisms appeared against Anthropic's decision. Some users questioned whether the watermarks could reveal the use of artificial intelligence in academic or professional work.
What criticisms did the new Claude system receive
One of the most critical posts was published by a user identified as visionode. According to their argument, regular users of Claude could be exposed while those familiar with other tools would have more chances to hide the origin of the content.
The user mentioned examples such as students using Claude to reorganize a paragraph, journalists summarizing lengthy transcripts, and writers turning to the chatbot to find synonyms.
Regular users of Claude could be exposed
What other Reddit users responded
The stance of visionode received critical responses from other Reddit users. Some questioned the argument and asked them to tone down their complaint.
Another user stated that the watermark does not seek to credit Claude, but rather to allow the detection of content generated by artificial intelligence.
A different criticism of the measure also emerged. One user described the decision to mark AI-generated content as ironic due to the origin of the data used to train the models.
What did other Reddit users respond?
What users who support Claude's watermarks say
Not all comments were negative. Some users considered the system a reasonable tool for identifying material generated by algorithms.
In one of the discussions, a user argued that there is no good argument against this type of identification. According to their view, the main reason to reject it would be to present AI-generated content as if it were not.
What users who support Claude's watermarks say
The debate on the use of artificial intelligence
The discussions around Claude focused on the possibility of detecting AI-generated content and the consequences of that identification.
While some users believe that watermarks allow for transparency in the use of these tools, others question how they could affect those using Claude for academic, journalistic, or professional tasks.
