X to Notify Users When Posts Are Shadowbanned Due to Legal Compliance

X on Monday announced it will now tell users when their posts have been “shadowbanned,” meaning shown to fewer people, due to the company’s compliance with local laws. The feature is an expansion of the social network’s earlier transparency tool called “Under the Hood,” which allows users to see if any labels have been applied to their account or their posts that could have affected their visibility on the platform.

It will also highlight any downranking of users’ posts due to government-required content filtering.

Previously, information about how an X user’s account or posts were being labeled and downranked was not available to end users. As a result, many often accused the company, formerly known as Twitter, of having shadowbanned them.

In addition, most companies that share data about government requests do so on an aggregate basis through “transparency reports,” not at the account or post level.

That began to change when, in August, X rolled out its transparency tool to offer more visibility into its moderation decisions. The “Under the Hood” option in the X app settings allows anyone who has posted at least 10 times during the month to download their aggregate stats as a JSON file, a standard data format. (Non-technical users can drop the file into any AI chatbot to have it explained.)

This same process will now show if an account or posts have had their visibility limited because X is complying with local laws. The file will also include which posts were withheld in a country following a legal demand, and which country made the request.

Governments cite a range of reasons for such demands, from laws against hate speech to terrorism-related content to election rules. Of course, critics, including free-expression groups, say some countries also use these laws to squash political dissent.

The update follows a number of other transparency-related efforts at X, including the open-sourcing of its “For You” algorithm, which selects the posts in users’ main feed, and an expansion of its efforts to open source parts of its codebase. The latter involved adding more detail, including on the model configuration, filtering, and core ranking system details. That means the code now includes the parameters used to weight different signals — information that is key to understanding which posts are actually displayed.