On August 15, Business Insider reported that Anthropic's approach of adding watermarks to text generated by Claude is prompting some users to abandon the AI assistant. Anthropic announced this week that certain versions of Claude will embed 'imperceptible watermarks' in output text. These watermarks will propagate when the text is copied and pasted, and may still remain after some editing, helping to distinguish between AI-generated and human-created content. Anthropic emphasized that the watermarks will not affect the meaning, quality, or readability of Claude's responses.