Twitch has released a technology that utilizes machine learning to detect persons attempting to re-enter chat channels from which they were previously banned due to abusive behavior.
"Bad actors" frequently established new accounts to abuse users, according to the gaming-focused livestreaming platform.
However, if a person was a "likely" or "potential" ban evader, the new algorithm would alert broadcasters and chat moderators.
It's part of Twitch's ongoing efforts to combat harassment and hate speech.
Hate raids, in which unethical broadcasters send their fans or even automated bots to other channels to attack someone, have been a source of criticism for the corporation.
Frequently, the victims are members of minority or marginalized communities.
Creators had pressed Amazon, which is owned by Amazon, to do more to combat hate speech.
Twitch unveiled "phone-verified chat" in September, allowing streamers to have certain or all users authenticate their phone numbers before conversing.
It also filed a lawsuit against anonymous individuals who were allegedly participating in "chat-based assaults against marginalized broadcasters" in the same month.
The new suspicious-user detection system is "powered by machine learning" and uses "a number of account signals" to detect ban evaders, Twitch said.
"Machine learning" describes computer systems that, in effect, "learn from experience".
By default, the new system will be switched on, but moderators and authors will be able to change its settings or turn it off.
It analyzes a variety of indicators, including the behavior and account attributes of players attempting to join a chat channel, to those of banned accounts, and flags potential ban evaders in two ways, according to Twitch.
- likely: in which case their chat messages will be blocked
- possible: in which case their messages will still appear
"No machine learning will ever be 100% accurate," Twitch said, so chat moderators would make the "final call", but the tool "will learn from the actions you take - and the accuracy of its predictions should improve over time".
