top of page

Twitch Is Using Machine Learning to Catch Ban-Evading Trolls

In its efforts to reduce harassment, Twitch has introduced Suspicious User Detection, a new tool that uses machine learning to detect people who attempt to evade bans. The new tool is helpful in combating hate raids, where streamers' chats are flooded with hateful messages.

The tool will be able to identify potential ban evaders by evaluating their behavior and characteristics about their accounts and further comparing the information with accounts that have been banned from a streamer’s channel.

While streamers will still be able to see messages sent by these "possible evaders", they won't be sent into the chatbox. Streamers can choose to monitor a "possible evader" and the user will be put into a monitoring list, or completely ban them.

According to Twitch, the Suspicious User Detection feature will be enabled by default but streamers can opt to disable it if they want. They can manually select suspicious users to be monitored.


Banner VBS16 EVO_728x90.jpg
bottom of page