How to Stop Spam in Discord

Most Discord spam isn't random — it's automated. Compromised accounts, scam bots, and coordinated campaigns post the same crypto-giveaway link, fake Nitro offer, or scam invite across many servers within minutes, often faster than a human moderator can react.

The larger or more public a server is, the more attractive it is to this kind of automation: a single working scam link can reach thousands of members before anyone reports it.

Discord ships a built-in AutoMod system that blocks messages matching keyword lists, mention-spam rules, and a small set of preset filters. It's a useful first layer, but it works on message content alone — it doesn't track patterns of behaviour across messages, doesn't inspect images, and has no way to know an account was already flagged as compromised somewhere else.

• Keyword and mention-spam filters catch obvious, static patterns
• Doesn't evaluate images, attachments, or behaviour across multiple messages
• No visibility into an account's history outside your own server

Anti-Scam runs continuously in the background on every Airwavy server — there's no command to turn on, and nothing to wait on a human to notice. It combines three independent detectors — message content, URLs, and behaviour over a rolling observation window (60 seconds by default) — so it can catch a burst of similar messages, not just one obviously-bad one.

• Free on every server: message, URL, and behaviour detection, each independently toggleable, with a configurable block-confidence threshold (default 70).
• A confirmed match can auto-delete the message and post a staff alert to a private channel you choose.
• Server Premium adds AI image-scam detection (catches scam screenshots text filters miss) and opt-in cross-server protection — the moment one server flags a compromised account, every opted-in server in the network can act on it too.

1. Add Airwavy to your server and open the dashboard.
2. Open Anti-Scam and choose a private staff channel — this is where alerts land, so pick one only staff can see.
3. Leave message, URL, and behaviour detection on (all three are enabled by default) and decide whether to auto-delete matches or have staff review them first.
4. Save, then watch the alert channel for the first few flags — adjust the block-confidence threshold if it's too aggressive or too lax for your community.
5. If you have Server Premium, also enable AI image-scam detection and decide whether to opt into cross-server compromised-account protection.

• Turning on auto-delete before reviewing a few alerts — start with staff review, then switch to auto-delete once you trust the threshold for your server.
• Pointing alerts at a channel the wrong people can see — flagged links are often still live; keep the alert channel staff-only.
• Assuming Anti-Scam covers every kind of spam — it targets scam-pattern content specifically. Off-topic flooding or harassment that isn't scam-shaped still needs a human call, or AI Moderation's broader signal set.
• Skipping AI Moderation — Anti-Scam is pattern-based; AI Moderation's 14 categories (Premium) catch cases without a clean pattern match, like ambiguous phrasing, mass mentions, or suspicious attachments, and only ever alerts or deletes, never bans.