Claude Watermark Removers Surge 60% as Anthropic’s AI Labels Trigger Privacy Fears
Updated
Updated · Gizmodo · Aug 19
Claude Watermark Removers Surge 60% as Anthropic’s AI Labels Trigger Privacy Fears
3 articles · Updated · Gizmodo · Aug 19
Summary
Google searches for “AI watermark remover” jumped 60% week over week after Anthropic began embedding imperceptible labels in all Claude-generated text earlier this month.
Removal tools are spreading fast: Guillaume Meyer’s open-source GitHub project has drawn more than 15,000 stars and thousands of forks, while other developers have launched web tools to scrub residual markers.
Users and toolmakers say the labels could wrongly imply an entire document was AI-written when only part was generated, and they worry watermarking could later be used in disputes over attribution or ownership.
Anthropic says the watermark only tests whether Claude may have produced or processed content and does not affect user rights, but the company adopted it to meet the EU AI Act’s output-marking requirement.
No rule currently bars users from stripping those labels, setting up a broader cat-and-mouse fight already familiar from AI image and video watermark removal.