Anthropic Reportedly Watermarks All Claude Outputs, Sparking Backlash

dreamwieber · x · 2026-08-12

A user recently highlighted that Anthropic is taking a heavy-handed approach by seemingly marking all text passing through Claude with a watermark.

The user criticized the policy for lacking distinction, failing to differentiate between text merely checked for grammar and content wholesale generated from scratch. This blanket approach could negatively impact many users who use Claude for routine assistance, potentially flagging their work in future AI detection systems.

Related event: Anthropic Introduces Invisible Watermarks for Claude to Comply with EU AI Act(10 posts)→

Original post →

More from Models

Models channel →