How Claude's text watermarking works

Summary

Anthropic describes a text-watermarking system planned for Claude to estimate whether Claude was involved in generating or processing text, primarily to comply with the EU AI Act. The method uses a key and preceding words to influence the source of randomness in low-stakes word choices, creating a detectable pattern without adding characters, tokens, identifying information, or extra cost. Anthropic says internal testing found no impact on content, creativity, readability, or perceived quality, while detection confidence improves with longer passages and weakens when text offers few valid alternatives, such as factual statements, proofreading, or code. The watermark can indicate only that Claude was likely involved, not whether text was human-written, who authored it, or which person or organization used Claude; light editing may not remove it, but a complete rewrite will.

Takeaways

1. 1

   Watermarking changes the source of randomness for low-stakes word choices rather than forcing Claude toward particular words or adding hidden characters.

2. 2

   Detection is less effective for short passages and text with few flexible choices, including factual statements, lightly edited human writing, proofreading, and code.

3. 3

   The detection API can indicate likely Claude involvement, but cannot establish human authorship, identify a person or organization, or distinguish writing from heavy editing.