Claude's Hidden Watermarks Spark Workplace Privacy Fears

Anthropic has started embedding invisible watermarks in Claude's text output, a change tied directly to the European Union's AI Act Transparency Code. The rule requires companies operating in the EU to mark AI-generated or AI-edited content in ways machines can detect, even if human readers can't spot anything unusual. Anthropic's rollout satisfies that mandate. It also means anyone using Claude for work, school, or writing now leaves a trace baked into the text itself, one that survives unless a user actively strips it out through paraphrasing tools or other cleanup.

That tradeoff has landed badly with a chunk of Claude's user base. Reddit threads on the change have filled with complaints from people who see the watermark as a trap for casual users rather than a transparency tool for regulators. The debate isn't unanimous.

Plenty of posters have pushed back on the loudest critics. But the loudest voices frame this as less about disclosure and more about who gets caught and who doesn't, depending on how much effort they put into hiding their AI use afterward. One recent post, from an account only three weeks old, laid out that argument in stark terms.