i personally think there are valid reasons to push back against watermarking imo: 1. the same watermarking techniques used to indicate that a text was generated by claude could be used to convey "this text was generated during a conversation with this specific user". i think it is reasonable to expect that governments will eventually require the watermarks to carry additional information in the future. 2. from a consumer perspective, "using people without giving them credit" is not what is going on here. you are paying anthropic in exchange for a service. would you be happy to find out that your plumber has decided to start writing secret messages in your wall saying "Jake was here"? also, the copyright status of AI generated works is not a fully settled question, and i personally do not want to risk Anthropic/OpenAI/Google having any legal ownership of my codebase because i occasionally used their model's assistance.