Twitch is launching a new machine learning feature called “Suspicious User Detection” that will automatically determine users it suspects might be trying to avoid a ban. The platform will block any messages that they send from appearing in chat.
Twitch’s Suspicious User Detection will identify ban evaders
This new Suspicious User Detection tool uses machine learning to identify suspicious users that may be using a new account. This way, streamers or mods will be able to take action against those suspicious accounts.
Suspicious User Detection will be open to anyone, but the tool won’t ban or block any account automatically. Because it’s impossible for a machine-learning algorithm to always make the best decision.
Twitch stated: “You’re the expert when it comes to your community, and you should make the final call on who can participate. The tool will learn from the actions you take and the accuracy of its predictions should improve over time as a result,” in its blog post.
Hate raids on Twitch have been a big problem for the platform. Thousands of hateful bots are used by attackers to post spam and other hate speech on Twitch chats. It looks like the platform is trying to find new security measures in order to prevent those attacks.