X users will soon have a way to actually back up their complaints about “shadowbans” affecting their reach on the service. The platform is experimenting with a new tool that gives users some details about how their posts are treated by the algorithm, including whether they have been been limited in some way.
The feature, which is currently only available to a “small group” of testers, is available via a new “under the hood” page in the app’s settings. People who have access to the tool can download data associated with their account and dig through it to find information “visibility-limiting labels” affecting their posts.
X announced the feature alongside an update that open-sources more aspects of the ranking algorithm that powers the “for you” timeline. X’s VP of Product Keith Coleman called the changes “an unprecedented level of transparency into the X algorithm.”
But actually using the feature might be a bit more complicated than you’d expect. To start, downloading data via “under the hood” doesn’t provide stats on a per-post basis. Instead, it “lets people see aggregate statistics” from the last month about “visibility-impacting labels on their account and posts,” according to details X shared on GitHub.
And the information it does provide could be hard to parse for some. An example, shared by @cb_doge (also known as Doge Designer) of what “under the hood” data looks like. In the video, the download results in a .json file that at first glance might look like a bunch of code. Look closely, and you can see that two of the account’s posts were hit with two separate “NSFW” labels that resulted in those posts being hidden from non-followers and underage users.
Screenshot via X
X has previously open-sourced versions of its recommendation algorithm, but those code dumps have revealed very little about how the platform actually works. But the latest release does shed give some insight into the type of content X deems ineligible for recommendations. For example, the GitHub repo lists numerous labels related to different types of NSFW posts as well as those for spam, violence, impersonation and hateful conduct. There are also more specialized labels, like one called “for emergency use only,” that’s described as an “emergency incident-response label with a custom notice.”
There do not appear to be any labels related to political content, though there is a “civic integrity” flag that restricts a post’s visibility to its author’s profile only. The label is used for posts “detected in response to a user report to contain content in violation of X’s Civic Integrity policy.” According to the company’s policy, this label is “reserved for critical escalations.”
In addition to “under the hood,” X said that “third-party recommender system experts” were able to preview and pressure-test this code release in advance.” It’s not clear how the company selected these experts or if their findings will be made public. Elon Musk recently said that “every line of code touching the X system will be open source and third-party audited” sometime in August.
The company’s latest release doesn’t exactly accomplish that. In fact, it notes that it continues to hold back some information that could potentially be used to “game the system” and manipulate the algorithm. X also hasn’t released the underlying code that deals with ads and other systems outside of its “for you” timelines. A spokesperson for X explained that future updates could include additional labels related to brand safety.
Update, August 13, 2026, 4:40PM PT: This post was updated to clarify that third-party experts have already reviewed X’s release and to add additional info from an X spokesperson.
