> ## Content Index
> Fetch the complete content index at: https://adjacent.media/llms.txt
> Use this file to discover other available public pages before exploring further.

# Anthropic's watermark distorts word choice, raising quality questions
- URL: https://adjacent.media/signals/anthropics-watermark-distorts-word-choice-raising-quality-questions/
- Published: 2026-08-17T16:08:55.000Z
- Updated: 2026-08-17T16:08:55.000Z
- Description: Anthropic claims its new text watermarking system—which subtly adjusts Claude’s word probability distributions to embed detection fingerprints—has no impact on output quality. John Gruber’s analysis suggests the mechanism biases model outputs away from optimal token selection.
- Author: Jonathan Greene
- Tags: #signal, theme-ai, model capability, output quality, safety/alignment

Source: [Daringfireball](https://daringfireball.net/2026/08/anthropics%5Fwatermark%5Ftext%5Fadulteration%5Fin%5Fclaude%5Fis%5Fa%5Fperversion%5Fof%5Fwriting?ref=adjacent.media)

Anthropic claims its new text watermarking system—which subtly adjusts Claude's word probability distributions to embed detection fingerprints—has no impact on output quality. John Gruber's analysis suggests the mechanism biases model outputs away from optimal token selection. This exposes a tension in AI safety infrastructure: detection methods that work probabilistically degrade the thing they're designed to protect, trading detection capability for measurable degradation that Anthropic hasn't quantified or disclosed.