Twitch Introduces New Tools to Combat Ban Evasions

Machine learning will flag potential accounts, but they'll still need moderation

Twitch has rolled out new Suspicious User Detection tools, which use machine learning to flag "likely" and "possible" accounts that are evading bans.

Plans for detecting ban evasions were announced back in August, and almost six months later, Twitch is making Suspicious User Detection available to all channels. According to the platform, the new tools should make it easier for streamers and moderators to catch and deal with new accounts from previously banned users.

Rear view of a gaming setup with desktop pc and a big monitor

mikkelwilliam / Getty Images

Twitch acknowledges that some users will create a new account after being banned from a channel in order to jump back into the chat and continue being abusive.

Suspicious User Detection utilizes machine learning to look for 'signals' that will help identify such accounts. After which the possible ban-evading accounts will be flagged, leaving it up to streamers and their mods to decide how to deal with them.

"Possible" ban evaders will still appear in chat, but the flag will be visible to the streamer and mods so they can keep an eye on them.

"Likely" evaders will also be flagged, but their messages won't appear in the public chat—only the streamer and mods will be able to see them. Channels can choose to hide messages from "possible" offenders as well, if desired.

"Likely" and "Possible" ban evading users


Suspicious User Detection is available for all Twitch channels now and is enabled by default. If desired, streamers can opt to turn the tools Off or make other adjustments.

It's also possible to manually add accounts to keep an eye on, either by using the Suspicious Users widget or directly from the user's Viewer Card.

Was this page helpful?