Techies have concerns about Claude's hidden watermark. Anthropic has some answers.

Anthropic, the developer of Claude, has unveiled a watermark for the AI tool to differentiate AI-generated content from human-written text. However, some tech professionals have raised concerns about the implications of this new feature. The watermark, an imperceptible mark embedded in Claude-generated text, is designed to travel with the content when copied or pasted, potentially surviving some editing.

This move is intended to comply with the European Union's AI Act and has been adopted by other AI companies such as Google and OpenAI. Critics, including tech investor Bill Gurley, have expressed concerns about the watermark potentially being used as a judge, jury, and prosecutor by Anthropic. They argue that only Anthropic would be able to identify the watermark, which could lead to biased decisions.

Some developers also worry that lightly edited AI content, such as grammar checks or translations, could be flagged as AI-authored, affecting the quality and readability of the responses. Additionally, there are concerns about digital privacy and data retention, as former Microsoft executive Steven Sinofsky pointed out. Nevertheless, some see potential benefits in watermarking, such as preventing AI systems from being trained on their own outputs and increasing transparency for AI-generated content.