Twitch strengthens methods to combat child predators

Twitch is continuing to build upon preventative measures against child predators on the platform, updating its policy to include mandatory phone verification for what it called "vulnerable accounts" alongside other steps to keep under-13s safe and away from the site, as they're prohibited from using Twitch under its existing policy.

"Keeping people safe online requires a constantly evolving approach. This is true of our big picture for safety on Twitch, but it’s especially true for a topic like child predation, where the potential harm is so appalling and unspeakable," Twitch said in its update today.

The update comes two months after Twitch initially highlighted methods it was using to combat predation. The changes came in response to a September Bloomberg report that explored the egregious ways child predators engaged with the site to view and manipulate children who were livestreaming, despite Twitch technically disallowing anyone under the age of 13 from using the site. Children between the ages of 14 and 17, by Twitch's rules, are allowed to stream and make an account on the site.

Bloomberg's reporting noted numerous instances where tens or hundreds of viewers watched underaged children, encouraging them to perform various tasks or acts in the chat. Through a researcher, Bloomberg was able to verify 1,976 accounts that followed numerous children, with the report stating that the site had discovered even more accounts that had not been catalogued by the researcher.

At the time, Twitch said it would "continue to aggressively expand the staff, technologies, industry relationships and internal processes" to stop children under 13 from using the site and prevent predators from exploiting anyone who slipped through the cracks. Now the platform has offered further explanation into what it's doing to prevent the problem growing.

Twitch clarified that "there is no single fix to prevent predation," and added that it couldn't share all its methods "to avoid giving bad actors information they could use to evade our efforts." But from now, the site is implementing mandatory phone verification requirements "Before potentially vulnerable accounts are able to livestream," plus blocking users previously suspended for being underage from making new accounts. Twitch also said it is "continuing to refine the moderation technology" used by staff to review reports. 

Default privacy settings for whispers have been changed, certain search terms to find streams have been blocked, and Twitch has acquired language processing artificial intelligence Spirit AI to aid "in building more nuanced classifiers for detecting harms of all kinds in written text on Twitch." These changes have all happened in the last two months, with the site continuing to work on preventative measures.

Twitch classifies child predation as "imminent harm," which falls under one of the site's most prioritised categories for threats. It says "collaboration is critical," encouraging anyone with information about this malicious activity on Twitch to share so that appropriate action can be taken.