S2. Ep32 - Anthropic Is Watermarking AI Text — And You Can't Strip It Out

AI Automations for Business · Aug 20, 2026 · 26m

There's a lot of misinformation flying around about Anthropic's new AI watermarking, so Noel breaks down what it actually is.

It isn't hidden code you can delete. The watermark is the words themselves and the statistical order they come out in, which means copying your text into a Google Doc and pasting it elsewhere won't shake it off. Noel explains how it works, why it's driven by EU law, which models it applies to and when, the API Anthropic is about to release so anyone can check whether text came from Claude, and what happened when he asked Claude to help him strip it out.

Also in a…