AI Moderation
Automatic detection of inappropriate content using artificial intelligence (OpenAI).
What does it detect?
| Category | Description | Action |
|---|---|---|
| CSAM (minors) | Child sexual abuse material | Permanent ban (not configurable) |
| NSFW | Explicit sexual content | Configurable |
| Scam/Phishing | Scams, fake investments | Configurable |
| Violence | Graphic violent content | Configurable |
| Harassment | Threats, harassment | Configurable |
| Self-harm | Self-harm content | Configurable |
How does it work?
- The user sends a message (text or photo)
- The message is delivered normally (without delay)
- In the background, the AI analyses the content
- If prohibited content is detected: the message is deleted and the action is triggered
Important: The analysis is asynchronous — the message may be visible for 1-2 seconds before being deleted.
Configuration
/config→ Select group → 🤖 AI Moderation- Enable/Disable — Toggle on/off
- Action — For non-CSAM categories:
delete,warn,mute,kick,ban
Zero tolerance for CSAM
CSAM detection (child sexual abuse material) has a very low threshold and always results in a permanent ban. This action is NOT configurable — it is a mandatory security measure.
Notes
- Administrators are exempt
- Analyses text and images (photos)
- Does not analyse videos, GIFs or stickers
- If the AI is unavailable, messages pass normally (fail-open)
- Admins receive a private notification when the AI takes action