# Improving AI spam detection for edits and merges
**URL:** https://meta.discourse.org/t/improving-ai-spam-detection-for-edits-and-merges/376898
**Category:** Feature
**Tags:** ai, spam
**Created:** [August 1, 2025, 1:34pm UTC](https://meta.discourse.org/t/improving-ai-spam-detection-for-edits-and-merges/376898 "2025-08-01T13:34:54Z")
**Posts on this page:** 2
**Page:** 1
### Author:  [@Max\_Salishchev](https://meta.discourse.org/u/Max_Salishchev)
#### Post date: [August 1, 2025, 1:34pm UTC](https://meta.discourse.org/t/improving-ai-spam-detection-for-edits-and-merges/376898/1 "2025-08-01T13:34:54Z")
Hi there,
We recently encountered an issue with the AI spam detection system. When posts from users with Trust Level 1 or lower were merged, moved, or edited, the AI flagged them as spam— even though they weren’t flagged when originally posted.
My colleague Helga spoke with someone from your Support Team, who explained:
_“The AI spam detection feature analyzes the first three posts from users with Trust Level 1 or lower, including edits. In this case, the tool was triggered because an edit was made to a new user’s first post.”_
Their recommendation was to add a custom instruction. I tried the following:
_“Do not flag posts from users with Trust Level 1 or lower if they were moved, merged, or edited by forum admins (Max and Helga).”_
Unfortunately, this didn’t resolve the issue. Could you advise on how to adjust the custom instruction to make it work properly?
Thanks for your help!
---
### Author:  [@Falco](https://meta.discourse.org/u/Falco)
#### Post date: [August 1, 2025, 4:48pm UTC](https://meta.discourse.org/t/improving-ai-spam-detection-for-edits-and-merges/376898/2 "2025-08-01T16:48:09Z")
> [@Max\_Salishchev](#):
>
> When posts from users with Trust Level 1 or lower were merged, moved, or edited, the AI flagged them as spam— even though they weren’t flagged when originally posted.
Was this a merge / move / edit by a staff user? If so, we should skip checking when the edits come from staff.