Anthropic's Text Watermark Is Easy to Bypass on Older Claude Models
Anthropic is rolling out text watermarks to four more Claude models before the EU AI Act's December deadline. But on older models like Claude Opus 4.6 and Sonnet 4.6, API users can switch off the watermark by changing a single setting. That makes the safeguard easy to bypass.
Quick glossaryAnthropic, text watermark, EU AI Act, API
- Anthropic
- An AI company based in San Francisco, best known for its Claude chatbot and the AI models behind it.
- text watermark
- A hidden signal added to AI-written text so that people or software can later spot that a machine wrote it.
- EU AI Act
- A European Union law that sets rules for how artificial intelligence can be built and used across its member countries.
- API
- A set of rules that lets software and developers connect to an AI model and use it directly in their own products.
Burning Potato score: 71%