Claude Refuses Watermark Removal, GLM Complies

kimmonismus · x · 2026-08-17

A user attempted to use Claude to install a GitHub tool designed to strip AI watermarks but was blocked by safety filters, citing Anthropic's policy and EU regulations. Despite arguments about customer consent and data privacy, Claude refused. The user noted that Chinese models like GLM 5.2 lack such restrictions and successfully performed the task, highlighting the contrast between Western safety compliance and uncensored model utility.

Related event: Claude Refuses to Install Watermark-Removal Plugin While GLM Complies, Sparking Safety Debate(2 posts)→

Original post →

More from Models

Models channel →