Anthropic adds watermarks to Claude outputs, and some users are mad it will expose cheating

Some Claude users are mad that Anthropic’s new watermarks will catch them using it at their jobs, classes

Anthropic now embeds invisible watermarks in Claude's text outputs to comply with the EU AI Act's transparency code. Complaints surfaced fast on Reddit and X: people worry that bosses or teachers will scan their work and catch them using AI for reports or assignments. The post doesn't explain how the watermark works technically, whether it can be stripped, or if Anthropic plans an opt-out for paying users.

Why it matters: Anthropic added invisible watermarks to Claude output, and Reddit/X users are furious about getting caught by bosses and teachers. Strong topic, but the post lacks technical details and user controls — just enough to hit the featured threshold.