In this week’s round up we cover X’s chat update and shifts in review approaches for Meta and YouTube. And, if you’re struggling with complex topics like money laundering and credit scores, TikTok has an AI cat who’s got your back.
Meta’s shift to AI reviewers
Meta is allegedly planning to swap out human risk reviewers for AI. Until recently, these “privacy and integrity reviews” were mostly handled by people. But now, internal documents suggest that up to 90% of those checks might soon be automated. This could mean a win for product teams who can roll out updates and new features faster. However, others are worried this shift could mean letting AI make important calls about potential real-world harm without enough human oversight. Meta said that “human expertise” is still being used for “novel and complex issues,” and that only “low-risk decisions” are being automated.
XChat has begun rolling out to beta testers, with some of X’s subscribed users reporting that they can access it. XChat is intended to be a more robust version of X’s existing DM (direct messages) feature, with features that could include group messages, end-to-end encryption, vanishing mode, the ability to mark messages as unread, and file sharing.
YouTube has updated its content review system to maximise the ad opportunities for creators. Currently, videos uploaded to YouTube are reviewed and deemed worthy of ads (green), some ads (yellow) or no ads (red). Creators can appeal the decision, but YouTube is making that easier with an automatic review for those who don’t quite make the cut. In their words: “To improve the accuracy of our yellow icon decisions and get your videos monetising faster, we are experimenting with automatically sending videos that receive a ‘Limited or no ads’ rating for an additional review that may be completed by a human”.
