How to Set Up Discord AutoMod to Block Bad Words, Spam, and Scam Links

To set up Discord AutoMod, open Server Settings then AutoMod, create rules under the "Block Custom Words" and "Block Spam Content" presets, add a keyword list, enable the suspicious-link and spam filters, and assign a response action like blocking the message or timing out the member. AutoMod is built into Discord, free, and works on every server, though some presets require enabling Community mode first.

Native AutoMod handles the obvious cases well. Where it struggles is the gray area: misspelled slurs, sarcasm, context that flips a "bad word" into a harmless one, and scammers who rephrase their pitch faster than you can update a blocklist. This guide walks through the full native setup, then shows exactly where it stops being enough.

What Discord AutoMod can and can't do natively

AutoMod is Discord's first-party content filter. It scans messages and nicknames the moment they're posted and acts before a human moderator ever sees them. It runs server-side, so it doesn't lag or go offline the way a third-party bot occasionally can.

What it does well:

• Keyword blocking from your own custom word list, including wildcards.
• Preset word lists Discord maintains for common profanity, slurs, and sexual content.
• Spam detection for repeated text and obvious flooding.
• Mention spam limits (for example, block a message that pings 10+ users or roles).
• Suspicious link flagging using Discord's own list of known scam and malware domains.

What it can't do:

• Read intent. AutoMod matches strings. It can't tell that "I'll kill this boss" is fine while "I'll kill you, meet me outside" is not.
• Adapt per channel. A word that's banned in #general might be normal in a #venting or #gaming channel, but AutoMod applies one rule everywhere unless you hand-build exemptions.
• Catch novel scam phrasing. It blocks known bad domains, but a brand-new "free Nitro" link or a redirect through a shortener often slips through until Discord's list catches up.
• Handle obfuscation reliably. Spacing out letters, swapping characters (l33t-speak), or using lookalike Unicode can dodge a plain keyword match.

That gap between "matches a string" and "understands what was meant" is the whole reason AI moderation exists, which we'll cover after the setup steps.

Step 1: Open Server Settings and AutoMod

Right-click your server name (or tap the server name at the top of the channel list on mobile) and choose Server Settings. In the left sidebar, scroll to the Moderation section and click AutoMod.

You'll see a list of rule presets:

• Block Custom Words
• Block Commonly Flagged Words
• Block Spam Content
• Block Mention Spam

If you don't see every preset, your server may need to be a Community server. Go to Server Settings then Enable Community and follow the short setup. You'll need the Manage Server permission to create or edit any AutoMod rule.

Before you build rules, decide who should be exempt. Almost always you'll want to exempt your moderator and admin roles so AutoMod doesn't block them mid-cleanup. You can set exempt roles and channels on each individual rule.

Step 2: Block commonly flagged words and a custom keyword list

You actually get two tools here, and you should use both.

Block Commonly Flagged Words uses Discord's maintained lists. Click it, then enable the categories you want:

• Profanity for general swearing.
• Sexual content for explicit terms.
• Slurs and hate speech for identity-based slurs.

These lists are updated by Discord and cover a lot of obfuscation automatically, so turn them on even if you also build a custom list.

Block Custom Words is your own list, and this is where you tailor moderation to your community. Click it, name the rule something clear like "Server-specific blocklist," and add your terms. A few rules of thumb:

• Use the wildcard * to catch variations. scam* matches "scammed," "scamming," and "scammer."
• Add common bypass spellings you actually see, since AutoMod won't always infer them. If members dodge a filter with "fr33," add "fr33."
• Add an allow list for false positives. Classic example: block *ass* but allow "class," "pass," and "assassin" so you don't nuke half your messages.
• Keep the list focused. A bloated blocklist causes more false positives than it prevents real abuse.

If you run multiple communities, write the list once and reuse it. The hard part is never the first ten words, it's maintaining the list as people invent new ways around it.

Step 3: Turn on spam and mention-spam filters

Back on the AutoMod page, open Block Spam Content. This catches obvious spam patterns: the same message posted repeatedly, mass-pasted text, and known spammy formatting. Enable it and set the response action (we cover actions in Step 5).

Then open Block Mention Spam. Set a threshold for the maximum number of unique user and role mentions allowed in a single message. A sensible default for most servers is 5. Raid bots love to mass-tag members to bait reactions, and a mention cap stops that without affecting normal conversation.

A few practical notes:

• Mention-spam blocking counts unique mentions, so a member quoting one person several times won't trip it.
• Pair this with Discord's built-in raid protection and verification level. AutoMod handles message content; raid protection handles the flood of joins. Both matter. Our ultimate guide to Discord server moderation covers how these layers fit together.
• Native spam detection is intentionally conservative to avoid false positives, so determined spammers who vary their text slightly can still get through. That's a known limit, not a setup mistake on your end.

Step 4: Enable suspicious and scam link blocking

Scam links are the highest-stakes thing AutoMod touches, because one "free Nitro" or fake-Steam-gift link can compromise dozens of accounts before a human notices.

Discord doesn't expose link blocking as a separate top-level preset on every server, so you handle it two ways:

1. Suspicious link detection. On servers where it's available, AutoMod can flag links to domains Discord knows are malicious. Enable it and set the action to block the message. This is your baseline.
2. Custom keyword rules for links. In your Block Custom Words rule, add patterns that catch the bait, not just the URL. Common scam phrasing includes free nitro, steamcommunity lookalikes, discord-gift, nitro-free, and shortener domains you don't want, like bit.ly if your community never uses them. Use wildcards: *discord-nitro* catches several variants.

The honest limitation: this is a blocklist. It catches what's already known. A scammer who registers a brand-new domain this morning, or rephrases the pitch, won't be on Discord's list yet and won't match your keywords until you add them. You're always one step behind a determined attacker. That's the exact gap an AI layer is built to close, covered in the next section.

Step 5: Choose the right response action per rule

Every AutoMod rule has a response action. Matching the action to the severity of the rule is what separates a clean server from one where moderators are constantly cleaning up overreactions.

Available actions:

• Block message. The message is stopped before anyone sees it. The author gets a notice. Use this for slurs, scam links, and spam, where the content should never appear.
• Send alert to a channel. AutoMod posts a copy of the flagged message to a private mod-log channel. Use this for borderline rules where you want eyes on it but not an automatic punishment.
• Timeout the member. Mutes the author for a set duration. Use this for repeat offenders or serious violations like slurs and scam links, on top of blocking the message.

A practical setup that works for most servers:

Rule	Action
Slurs / hate speech	Block + timeout + alert mod channel
Scam / suspicious links	Block + timeout
Custom blocklist (general profanity)	Block only
Spam content	Block + alert
Mention spam	Block + timeout

Always route alerts to a dedicated #mod-log channel that only staff can see. Reviewing what AutoMod catches tells you which rules are too aggressive (lots of false positives to allow-list) and which are too loose (real abuse slipping through). For a broader workflow on running moderation day to day, see how to moderate a Discord server.

Where native AutoMod falls short and how PeakBot's AI layer fills the gap

You can build a strong native AutoMod setup with the five steps above, and you should. But every weakness in this guide traces back to the same root cause: AutoMod matches strings, it doesn't understand meaning.

That shows up as:

• False positives. "I'll kill this raid boss" gets blocked because "kill" is on a list. Your members get punished for normal talk.
• False negatives. "you're a piece of human garbage, end yourself" sails through because none of those exact words are on your blocklist. The intent is obviously abusive; the string match is clean.
• Blocklist maintenance. Every new slang spelling, every new scam phrasing, every new bypass trick is a manual update. You're maintaining a list forever.
• One rule everywhere. Tuning per-channel exemptions by hand is tedious and easy to get wrong.

PeakBot approaches moderation differently. Its AI moderation reads the intent of a message and adapts per channel, instead of checking it against a fixed keyword blocklist. A threat phrased in words that aren't on any list still gets caught, because the model understands what was meant, not just which characters were typed. A "bad word" used harmlessly in context is left alone, which cuts the false positives that make members resent automated moderation.

This is the layer that sits on top of, not instead of, native AutoMod. Keep AutoMod doing the cheap, fast, known-bad blocking. Let AI handle the gray area AutoMod was never designed for: rephrased scams, obfuscated slurs, and context-dependent calls.

A few things worth knowing about PeakBot specifically:

• It's a free, AI-powered Discord bot. AI moderation is part of the 30+ features that are free with no time limit and no trial period.
• It also covers anti-raid and anti-nuke protection, full logging, and the rest of a moderation stack, so you're not stitching five bots together.
• It replaces MEE6, Carl-bot, Dyno, and TidyCord with one bot, and it currently powers 500+ Discord communities.
• Pro is $8.25/month (or $69/year) per server if you later want the AI Server Builder and other Pro features, but moderation itself doesn't require Pro.

To be fair to the alternatives: MEE6 has a polished dashboard and a huge install base, Carl-bot's reaction roles and automod logging are genuinely strong, and Dyno is cheap and reliable for basic filtering at $4.99/month. If you only want classic keyword automod, any of them works. PeakBot's edge is the AI layer doing what a blocklist structurally can't, plus bundling the rest of the toolkit for free. You can see the full feature list on the features page.

FAQ: AutoMod setup questions

Do I need Discord AutoMod and a bot, or just one?

Use both. Native AutoMod is the fast, free first line that blocks known-bad words, spam, and flagged links server-side. An AI moderation bot like PeakBot adds the contextual layer that catches rephrased threats and scams a static blocklist misses. They complement each other rather than compete.

Why does AutoMod block normal messages sometimes?

Because it matches text strings, not meaning. A word on your blocklist will be caught even inside an innocent word or sentence. Fix it by adding an allow list to your custom-word rule (for example, allow "class" while blocking "ass") and by reviewing your mod-log to find which terms cause the most false positives.

How do I stop scam and Nitro links if AutoMod misses them?

Enable suspicious-link detection where it's available, then add custom keyword rules for common scam phrasing like free nitro and discord-gift with wildcards. Because this is still a blocklist that only catches known patterns, pair it with an AI moderation layer that reads intent, so brand-new scam phrasing gets flagged before it's on any list.

Does AutoMod work on small servers?

Yes. AutoMod is available on any server where you have the Manage Server permission, though some presets require enabling Community mode first. It runs the same way regardless of member count, so even a small server gets full content filtering.

Is setting up AutoMod free?

Completely free. AutoMod is built into Discord at no cost. Adding PeakBot's AI moderation on top is also free, with no trial or time limit, so you can run a full moderation stack without paying for anything unless you later want Pro features.