# How can some sites use AI to bypass AI detectors by rewriting everything without losing context, but AI LLM's fail to do so, even if u strictly ask them to do so? — retained record

> Retained reddit record from reddit:ClaudeAI.

Canonical: https://highsignal.app/data/records/115bca8504418ae3

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:ClaudeAI)  published 2026-10-02 09:35 ingested 2026-10-03 02:36

[original source ↗](https://www.reddit.com/r/ClaudeAI/comments/1wvpfyc/how_can_some_sites_use_ai_to_bypass_ai_detectors/)

## retained text



```
Makes no sense that there are plenty of sites (usually paid) that can bypass those AI detectors, but big LLM's like Claude and ChatGPT can't? Even if u ask them strictly to do so? People claim you can't use AI to act like a human/not AI, but those sites manage to do so, meanwhile if you give Claude the exact humanizing structure, it still fails?
```



document fetched 2026-10-03 00:22 · canonical https://reddit.com/r/ClaudeAI/comments/1wvpfyc/how_can_some_sites_use_ai_to_bypass_ai_detectors
