Jukan: Anthropic adding watermarks on the eve of IPO is a serious self-harm

According to the analyst Jukan, the addition of an invisible watermark to Claude ' s model by Anthropic on the eve of IPO was a serious self-harm decision. Anthropic began adding machine-identifiable invisible watermarks to the text produced by the support models Claude, Claude Code, Cowork, API and AWS, Google Cloud, Microsoft Foundation. New models are supported online and existing models are added. In order to comply with the transparency requirements of Article 50 of the EU AI Act, Anthropic embeds the watermark directly into the text and the watermark can be glued with reproduction Dissemination and global application do not affect readability。