# Anthropic says new Claude models add output watermarks

_Tuesday, September 1, 2026 at 8:00 AM EDT · AI · Latest · Tier 2 — Notable_

![Anthropic says new Claude models add output watermarks — Primary](https://www.pcworld.com/wp-content/uploads/2026/09/3224834-0-85381600-1788293316-Claude-app.jpg?quality=50&strip=all&w=1024)

Anthropic's newly released Claude Fable 5.1 and restricted-access Claude Mythos 5.1 add invisible watermarks to text and file outputs, according to the company. Anthropic says the watermark is detectable through an API available to regulators, law enforcement, media, fact-checkers, independent researchers and other eligible groups. The company says its system steers word choice during inference and is intended to survive copying and light edits. Anthropic says an absent watermark does not prove text was human-written.

## Sources

- [PCWorld](https://www.pcworld.com/article/3224834/claudes-updated-fable-and-mythos-models-watermark-their-replies.html)

---
Canonical: https://techandbusiness.org/newswire/MIccDGdbPZUg-TUpbYtjZ-
Retrieved: 2026-09-01T23:03:15.184Z
Publisher: Tech & Business (techandbusiness.org)