Setting up AutoMod on Twitch takes about five minutes, and it is one of the highest-leverage things you can do for your channel in 2026. I have configured AutoMod on dozens of channels ranging from small gaming streams to large community broadcasters, and the difference in chat quality is always noticeable within the first week.
AutoMod is Twitch’s built-in AI moderation tool that holds suspicious messages for review before they appear in your chat. It uses machine learning trained on millions of chat messages, so it can catch obvious slurs, aggressive language, and creative misspellings that older keyword filters missed. In this guide, I will walk you through exactly how to set up AutoMod on Twitch, explain what each filter level means, and share the settings I recommend for different channel types.
Table of Contents
- Quick Steps: Set Up AutoMod on Twitch
- What Is AutoMod and Why Streamers Need It
- How AutoMod Works Behind the Scenes
- How to Set Up AutoMod on Twitch (Step-by-Step)
- AutoMod Filter Levels Explained (0 to 4)
- AutoMod Moderation Categories
- Configuring Blocked Terms and Permitted Terms
- How to Set Up AutoMod on Twitch Mobile
- Best AutoMod Settings by Channel Type
- Common AutoMod Problems and Fixes
- How to Turn Off AutoMod on Twitch
- Frequently Asked Questions
- Conclusion
Quick Steps: Set Up AutoMod on Twitch
If you want the short version, here is the entire setup process in four steps. You can complete this in roughly five minutes from your Creator Dashboard.
- Open your Creator Dashboard and click Settings in the left sidebar.
- Select Moderation, then scroll to the AutoMod & Message Filtering section.
- Toggle AutoMod on and pick a preset level (Level 1 is the default) or use the custom slider.
- Optionally add blocked terms, permitted terms, and adjust the four category sliders before saving.
That is the core process. Below I unpack every screen, every level, and every setting so you can fine-tune AutoMod for your specific community.
What Is AutoMod and Why Streamers Need It
AutoMod is Twitch’s first-line chat moderation system that runs 24/7, even when you are offline or AFK. Think of it as a tireless moderator who only flags suspicious messages and never makes a final call on its own.
The key thing to understand is that AutoMod does not ban or timeout anyone. It holds messages it considers risky in a review queue so your human moderators can approve or deny them. This hold-and-review model is what separates AutoMod from a hard ban filter and is the reason Twitch chose a moderator-in-the-loop design.
Every streamer benefits from AutoMod, but I have seen the biggest impact on three specific groups:
- New streamers without a mod team. You simply cannot watch chat, play a game, and engage viewers at the same time. AutoMod catches the worst messages so you can focus on content.
- Part-time streamers. When you go offline, your chat does not. AutoMod keeps the conversation clean during those gaps.
- Family-friendly and brand-safe channels. If advertisers or sponsors are watching, a properly tuned AutoMod is your insurance policy.
AutoMod is supported in 17 languages, which means it works well for international communities. It also learns from your moderators’ decisions. Every time a mod approves or denies a held message, the system gets smarter about your specific community’s norms.
How AutoMod Works Behind the Scenes
AutoMod is powered by machine learning and natural language processing models trained on Twitch chat data. When a user sends a message, AutoMod evaluates it against four content categories at the same time: identity-based language, sexual language, aggressive speech, and profanity.
The model is not just matching exact words. It understands context, common misspellings, and creative evasion attempts like accented characters, leetspeak, or inserted symbols. A message like “y0ur m0m” or “îdiot” is still flagged because the model normalizes text before scoring.
There is also a feedback loop. When your moderators approve or deny held messages, those decisions feed back into how AutoMod scores future messages from your community. Over time, AutoMod adapts to your specific chat culture, which is why I always tell streamers to wait at least two weeks before changing the level.
How to Set Up AutoMod on Twitch (Step-by-Step)
This is the full desktop walkthrough. I recommend doing the initial setup on a computer because the layout is clearer, then mirroring it on mobile.
Step 1: Open Your Creator Dashboard
Log into Twitch and click your profile icon in the top-right corner. From the dropdown, select Creator Dashboard. This is the control center for everything related to running your channel.
Step 2: Navigate to Settings, Then Moderation
In the left sidebar, click Settings. A submenu will appear. Click Moderation under that submenu. You will land on the Moderation settings page, which contains AutoMod, message filtering, blocked terms, and chat rules.
Step 3: Turn On AutoMod and Choose a Level
Scroll to the section labeled AutoMod & Message Filtering. You will see an AutoMod toggle at the top. Flip it on. Once enabled, choose from one of four preset levels (0 through 4) or click Choose level to open the custom slider.
Step 4: Adjust Each Category Slider
The custom slider lets you set a different intensity for each of the four categories: identity language, sexual language, aggressive speech, and profanity. Each slider runs from 0 (no filtering) to 4 (strict filtering). I recommend starting at Level 1 or 2 across all categories and adjusting after a week of observation.
Step 5: Save Your Settings
AutoMod saves automatically in most cases, but scroll to the bottom of the page to confirm. You should see a confirmation banner or a green checkmark. Your AutoMod is now active.
The whole process takes me about three minutes on a fresh channel. If you are migrating from an existing setup, give yourself five to ten minutes to update each category individually.
AutoMod Filter Levels Explained (0 to 4)
AutoMod levels control how aggressively the model holds messages for review. Each level builds on the one below it, so Level 4 catches everything Level 3 catches plus more. Here is the breakdown I share with every streamer I work with.
- Level 0 (Off): AutoMod is disabled. Messages are not held. Choose this only if you use a third-party bot that handles all moderation.
- Level 1 (Minimal): Catches the most obvious slurs, hate speech, and severe profanity. Safe for channels with an active mod team. This is my default recommendation for new streamers.
- Level 2 (Moderate): The Twitch default. Catches most slurs, sexual language, and aggressive content. Reddit users report this level flags roughly 95% of messages in busy chats, which is why I usually start streamers at Level 1 and work up.
- Level 3 (Strict): Catches borderline messages and creative spellings. Best for family-friendly channels or large communities where moderation cannot keep up.
- Level 4 (Maximum): The strictest setting. Blocks most messages containing any flagged patterns. Use this only for channels aimed at young audiences or strict brand-safe environments.
| Level | Identity | Sexual | Aggressive | Profanity | Best For |
|---|---|---|---|---|---|
| 0 | Off | Off | Off | Off | Channels with custom bots only |
| 1 | Low | Low | Low | Low | New streamers, active mod teams |
| 2 | Moderate | Moderate | Moderate | Moderate | Default Twitch setting |
| 3 | High | High | High | High | Family-friendly, large channels |
| 4 | Maximum | Maximum | Maximum | Maximum | Brand-safe, young audiences |
AutoMod Moderation Categories
Each AutoMod category controls a specific type of language. You can set these independently in the custom slider, which is what I recommend for most channels.
Identity Language
Flags slurs and offensive language targeting race, gender, sexuality, religion, or disability. This is the category I keep at Level 2 or higher regardless of channel type, because identity-based attacks are never acceptable.
Sexual Language
Flags explicit or suggestive sexual content. I set this to Level 1 for adult-friendly channels and Level 3 for family-friendly channels. Twitch’s Terms of Service prohibit explicit sexual content in chat regardless, so this category is mostly a reinforcement.
Aggressive Speech
Flags threats, harassment, and hostile language. I keep this at Level 2 for most channels. Aggressive speech is the second most common reason I see streamers lose viewers, and AutoMod is very good at catching it.
Profanity
Flags curse words and crude language. This is the most subjective category. For gaming channels, I often set this to Level 1 to allow casual swearing. For family-friendly channels, I push it to Level 3.
Configuring Blocked Terms and Permitted Terms
AutoMod works well on its own, but custom word lists are where you tailor it to your community. You can find these settings right below the AutoMod level selector on the same Moderation page.
Blocked Terms
Add any word, phrase, or pattern you want AutoMod to flag automatically. Common examples include competitor names (for sponsored channels), specific slurs used by trolls in your community, or spam URLs. Each blocked term can be marked as private, meaning it is visible only to you and your moderators, not to the chatters who try to use it.
Permitted Terms
This is the whitelist. If AutoMod is flagging a word it should not (like a character name or inside joke), add it to the permitted terms list. AutoMod will skip that word during scoring. I use this constantly for streamer usernames, game-specific terms, and community inside jokes.
Private Blocked Terms
The private blocked terms feature is one of my favorites. When you add a blocked term and mark it private, the word does not appear in any public message that explains why a chat message was held. This prevents trolls from reverse-engineering your filter list.
You can add up to 100 blocked terms and 100 permitted terms per channel. If you need more, you are probably better off using a third-party bot like Moobot or Nightbot alongside AutoMod.
How to Set Up AutoMod on Twitch Mobile
Setting up AutoMod from the Twitch mobile app is also possible, though the screen is smaller. I use the mobile setup when I am away from my desk and need to make a quick adjustment.
- Open the Twitch app and tap your profile icon in the top-left corner.
- Tap Creator Dashboard, then tap the settings gear icon in the top-right.
- Select Moderation from the settings menu.
- Toggle AutoMod on and choose a level from the dropdown.
- Scroll down to edit blocked terms and permitted terms.
One thing to note: the custom slider for individual categories is currently easier to adjust on desktop. I configure the category sliders on my computer and use mobile only for quick level changes or to add blocked terms on the fly.
Best AutoMod Settings by Channel Type
There is no single “best” AutoMod configuration because every community is different. After testing across dozens of channels, here are the presets I recommend as starting points.
Family-Friendly Channels
Set AutoMod to Level 3 or 4 with all four categories at their highest. Add common slang, innuendo, and dated memes to your blocked terms. This is the strictest configuration I recommend, and it works well for channels aimed at younger audiences or advertisers.
Gaming Channels
Set AutoMod to Level 2 globally, but drop the profanity category to Level 1. Gamers swear, and over-filtering profanity hurts chat engagement. Keep identity language and aggressive speech at Level 2 or higher.
Variety Streamers
Set AutoMod to Level 2 with profanity at Level 1. Variety streamers tend to have more diverse audiences, so a balanced configuration works best. Add game-specific terms to your permitted list so AutoMod does not flag character names or tournament jargon.
IRL and Just Chatting
Set AutoMod to Level 2 across all categories. IRL chat tends to be more conversational, which means AutoMod flags more messages by default. I usually start streamers at Level 1 here and let the moderator feedback loop tune the system for two weeks before increasing.
These are starting points. After running for two to four weeks, review your held messages queue and adjust the category sliders based on what is actually being flagged.
Common AutoMod Problems and Fixes
Even with a perfect setup, you will run into AutoMod issues. Here are the three problems I see most often, plus the fixes that actually work.
AutoMod Is Over-Flagging Normal Messages
This is the most common complaint. Reddit users regularly report that Level 2 catches 95% of their chat, including normal greetings like “hey” and “yo.” The fix is simple: lower the level to 1 or use the custom slider to reduce the aggressiveness of individual categories. If you want to keep the default level, add the over-flagged words to your permitted terms list.
AutoMod Is Not Catching Enough
If trolls are getting through AutoMod, you likely need to increase the level or add specific blocked terms. I also recommend reviewing your moderator team’s workflow to make sure they are actively approving and denying held messages. The feedback loop only works if your mods are using it.
AutoMod Conflicts With Third-Party Bots
Nightbot, Moobot, and Streamlabs can sometimes conflict with AutoMod. The most common issue is verified bots getting flagged by AutoMod. The fix is to add the bot’s username to your permitted terms list or use the bot’s verification feature to mark it as trusted.
If you run a bot like Moobot, you can configure it to automatically deny AutoMod-held messages, which removes the moderation bottleneck entirely. This is what I recommend for channels with a non-human moderator team.
How to Turn Off AutoMod on Twitch
To turn off AutoMod, go to your Creator Dashboard, click Settings, then Moderation, and toggle AutoMod off. The change takes effect immediately. You can also set AutoMod to Level 0, which is functionally the same. I only recommend disabling AutoMod if you have a third-party bot handling all moderation or if you want a completely unfiltered chat.
Frequently Asked Questions
What is AutoMod on Twitch?
AutoMod is Twitch’s built-in AI moderation tool that uses machine learning to hold suspicious chat messages for moderator review before they appear publicly. It runs 24/7 across four content categories: identity-based language, sexual language, aggressive speech, and profanity.
Can AutoMod ban users on Twitch?
No. AutoMod does not ban, timeout, or permanently remove users. It holds messages in a review queue so that your moderators can approve or deny them. Only a human moderator can issue a ban or timeout.
How many AutoMod levels are there?
AutoMod has five levels: 0 (off), 1 (minimal), 2 (moderate, the default), 3 (strict), and 4 (maximum). Each level increases how aggressively AutoMod holds messages for review. You can also use the custom slider to set each category independently.
Should I turn on AutoMod on Twitch?
Yes, in most cases. AutoMod is a low-effort way to enforce baseline chat rules, especially if you do not have a 24/7 moderator team. It is supported in 17 languages and adapts to your community’s norms over time through the moderator feedback loop.
Does AutoMod catch misspellings and slang?
Yes. AutoMod uses natural language processing and text normalization, so it catches common misspellings, accented characters, leetspeak, and other evasion attempts. The model is updated regularly based on moderator feedback.
Can moderators change AutoMod settings?
Channel moderators can review and approve or deny held messages, but only the broadcaster can change AutoMod levels, blocked terms, and permitted terms. This restriction prevents unauthorized changes to your moderation baseline.
Conclusion
Learning how to set up AutoMod on Twitch is one of the highest-leverage things you can do as a streamer in 2026. The core process takes five minutes: open the Creator Dashboard, go to Settings then Moderation, toggle AutoMod on, and choose a level. From there, spend ten minutes adjusting the four category sliders and adding a few blocked and permitted terms to match your community.
I recommend starting at Level 1 or 2, running for two weeks, and then reviewing your held messages queue. Adjust the category sliders based on what is actually being flagged rather than guessing. With this approach, you will have a clean, friendly chat that lets you focus on streaming instead of moderating.