How AI text watermarking works
A visual, interactive explainer of how statistical watermarks hide inside AI-generated text, and what erases them. It covers the core mechanism (secret-keyed nudging of word choices), how detection works via counting, the effect of editing, and practical implications. Includes demos and references to real systems like Google's SynthID and Anthropic's Claude.
They're invisible, they survive copying, and they work because they don't live in the characters at all. They live in the choices between them.