# Embedding watermarks in AI-generated text

**URL:** <https://boards.straightdope.com/t/embedding-watermarks-in-ai-generated-text/1032309>\
**Category:** Factual Questions\
**Created:** [August 12, 2026, 8:54am UTC](https://boards.straightdope.com/t/embedding-watermarks-in-ai-generated-text/1032309 "2026-08-12T08:54:49Z")\
**Posts on this page:** 3\
**Page:** 5

<div class="post-metadata">

**Author:** ![md-2000](https://avatars.discourse-cdn.com/v4/letter/m/9d8465/32.png) [@md-2000](https://boards.straightdope.com/u/md-2000)\
**Post date:** [October 12, 2026, 3:41pm UTC](https://boards.straightdope.com/t/embedding-watermarks-in-ai-generated-text/1032309/81 "2026-10-12T15:41:46Z")

</div>

> [@Chessic\_Sense](#):
>
> A watermark is like Alice identifying Bob by his dental records - it’s meant to work no matter what “they” did to Bob.

Unless they pull one or more teeth… 😃

---

<div class="post-metadata">

**Author:** ![Reply](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/reply/32/15952_2.png) [@Reply](https://boards.straightdope.com/u/Reply)\
**Post date:** [October 13, 2026, 2:44am UTC](https://boards.straightdope.com/t/embedding-watermarks-in-ai-generated-text/1032309/82 "2026-10-13T02:44:40Z")

</div>

So in images, there is a thing called a “perceptual hash”: [Perceptual hashing - Wikipedia](https://en.wikipedia.org/wiki/Perceptual_hashing)

Basically it compares “does this image look like this other one”, despite small modifications, crops, recoloring, etc. It detects similarity, not sameness.

I wonder if there’s something like that for text…?

---

<div class="post-metadata">

**Author:** ![CaveMike](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/cavemike/32/16379_2.png) [@CaveMike](https://boards.straightdope.com/u/CaveMike)\
**Post date:** [October 13, 2026, 3:19am UTC](https://boards.straightdope.com/t/embedding-watermarks-in-ai-generated-text/1032309/83 "2026-10-13T03:19:31Z")

</div>

> [@Reply](#):
>
> I wonder if there’s something like that for text…?

Text embeddings (e.g. BERT) will do this. Also not a hash algorithm, but there are algorithms like [Levenshtein distance](https://en.wikipedia.org/wiki/Levenshtein_distance).

[Previous page](https://boards.straightdope.com/t/embedding-watermarks-in-ai-generated-text/1032309.md?page=4)