Anthropic's New Watermarks Spark Backlash: Users Fear Job and Class Detection

They are invisible metadata embedded in Claude's generated text to identify it as AI-produced, making it easier to detect AI use.

Because they could reveal that users relied on Claude for tasks that may be prohibited or frowned upon, such as completing homework or work assignments, raising privacy and accountability concerns.

It could push companies and universities to establish clearer policies on responsible AI use, leveraging this technology to ensure academic and professional integrity.