The release, shared under an open-source license, allows developers to inspect the systems that filter content and even run parts of the ranking software independently.
This move is intended to address long-standing concerns regarding "shadowbanning"—the practice of making a user's posts invisible or undiscoverable without their knowledge.
By providing a new "Under the Hood" transparency tool, X will allow certain users to download a data file showing if any restrictive labels have been applied to their accounts or posts over the past month.
The company aims for this level of disclosure to create a "level playing field" that can be audited and critiqued by external researchers and the general public to verify the platform's fairness.
The initial rollout of the transparency tool is limited to a pilot group of accounts that are at least one year old and have posted frequently in the last month.
While most of the ranking logic is now public, X stated it has withheld specific systems, such as those using its Grok AI to predict rule violations, to prevent bad actors from circumventing spam filters.
Moving forward, the company plans to consider code contributions from the public to improve the algorithm, though it remains less transparent about other business metrics like revenue and government takedown requests.