Twitch recently unveiled a new AI tool that can help detect banned users trying to re-enter chat channels.<\/p>\n\n\n\n
As part of the company’s dedicated efforts to combat hate on the platform, the new system will alert streamers and chat moderators when a particular user “likely” or “possible” circumvents a ban.<\/p>\n\n\n\n
This tool will therefore restrict the opening of new accounts after they have been banned for hate and harassment offences.<\/p>\n\n\n\n
The new feature, known as the suspicious-user detection system uses “machine learning” and “a number of account signals” to detect ban evaders.<\/p>\n\n\n\n
Although it is enabled by default, moderators or creators can adjust, use, or switch it off.<\/p>\n\n\n\n
Although it has been categorically stated that machine learning is not 100% accurate, chat moderators have the final say on whether or not to use the feature.<\/p>\n\n\n\n
However, the company said that the tool “will learn from the actions you take – and the accuracy of its predictions should improve over time.” The new feature follows other implemented decisions put in place to combat hate including phone-verified chat and legal actions against unidentified users allegedly said to be involved in “chat-based attacks against marginalized streamers.”<\/p>\n\n\n