Anthropic last week announced that Claude would begin including hidden watermarks on text generated by its artificial intelligence service – in line with a request from the European Union. The move has also led to the discomfort of many users who rely on the service.
On social media, many users expressed their displeasure with the move, and have begun to deactivate their subscriptions to Claude’s service.
The watermarks included by Claude are invisible to the human eye, but can be detected by machines when the text is copied, and remain even after human editing. Some users have said they are uncomfortable with the presence of the mark, even after they have edited the text.
It is currently unknown how big an impact this will have on Anthropic’s business. This may also prompt other providers such as OpenAI to consider implementing similar measures.
