X will use Grok AI to better detect stolen content, redirect payouts to original creators, and crack down on engagement bait.

X is taking a tougher stance against people who try to abuse its creator revenue-sharing program by soliciting engagement and stealing content from others. It’s common for viral content to be recirculated, especially on platforms like X, where that content can be rewarded with likes, follows, and, in some cases, cold, hard cash. Instagram, Facebook, and Reddit also deal with this problem, and have implemented some technical measures to discourage such behavior — for instance, tools that detect when a user has reposted someone else’s work without giving them credit. X recently tried its hand at this, too, adding an improved video editor and recorder to its platform in an effort to encourage creators to post using X’s own tools instead of stealing others’ content. Now, Nikita Bier said X’s newest version of the Grok AI model can detect duplicated content at thrice the rate of its previous version. Plus, adding watermarks, intros, and other edits meant to fool people into thinking stolen content is your own will result in sending the monetized impressions to the original uploader. This will also include copies of viral text posts. (Bier says the most common of this kind is, “Twitter is like the smoking section of the internet” — proof that people are still calling the app Twitter!) Bier said the company has detected 1.5 million posts that were stolen in its latest cycle, though he didn’t mention over what period that cycle took place. He noted, however, that with the changes, over $1 million in creator payouts will now be given back to the original creators of the stolen content. Because many such problems are enabled by AI, X has been working to suspend bots more quickly, too. In April, Bier said the platform was identifying and suspending “208 bots per minute and growing.” X now says repeated or intentional attempts to circumvent its new policy will result in users being removed from the creator program, as will efforts to solicit engagement and follows. In terms of the latter, if a user is busted three times or more (e.g., by saying something like “I’ll follow everyone who replies”), their account will be removed from the creator program and forwarded to the platform’s policy team for suspension.