X opens up its For You algorithm and adds new tools to show how posts are ranked
X is taking a major step toward algorithmic transparency by expanding access to the source code behind its For You feed and core ranking systems. The social platform announced that it is making a much larger portion of its recommendation technology available to the public, giving developers, researchers, and users a closer look at how posts are selected, filtered, scored, and displayed.
The For You timeline is the default feed users see when they open X. It plays a central role in shaping what content gets attention on the platform, from breaking news and political commentary to entertainment, memes, sports, and creator posts. Because of that influence, users have long questioned how the system decides what appears in their feed and whether certain accounts or topics are being boosted, limited, or filtered.
With this latest move, X says it is publishing the source code for the For You timeline under the Apache v2 license. The company is also adding far more detail than in previous open source releases, including model configurations, filtering systems, and important parts of the core ranking engine. According to X, the newly released codebase is now roughly 10 to 15 times larger than before.
One of the most important additions is the inclusion of ranking parameters. These parameters help determine how different signals are weighted when the platform decides which posts to show a user. In simple terms, the update gives outsiders a better chance to understand why some posts rise to the top while others receive less visibility.
X says the newly opened code includes the main ranking logic that gathers posts for a specific user, scores them, orders them, and assembles the final feed. It also includes systems designed to filter out potentially problematic content, such as posts that may violate platform rules or create a poor user experience.
Some of these systems can even be run outside the company, giving researchers and developers the ability to study parts of the ranking process more directly. That could make it easier for independent observers to examine how X’s recommendation engine works in practice, instead of relying only on company explanations.
Alongside the expanded open source release, X is also introducing a user-facing transparency feature. The new tool will appear in an “Under the Hood” section inside the app’s settings. It is designed to help users see whether their account or posts have been affected by X’s ranking systems.
Users who have posted at least 10 times in the past month will be able to download aggregate account data as a JSON file. This file will show whether any labels were applied to their account or posts during the previous calendar month. These labels may indicate that a post was handled differently by ranking or moderation-related systems.
For technically skilled users, the JSON file can be reviewed directly. For everyone else, the information can be copied into an AI assistant and compared with X’s public codebase to help explain what the data may mean. This gives ordinary users a new way to investigate whether their content has been limited, classified, or affected by automated systems.
At launch, the transparency tool will be available only to a pilot group. X says the first test will include accounts that are at least one year old, with plans to expand access more broadly after the initial rollout.
Before announcing the update publicly, X previewed the expanded codebase to outside researchers with experience in recommendation systems. These researchers were able to train and run X’s Phoenix scoring system using the open source code. X views that as a significant milestone in its effort to make the platform’s ranking technology easier to inspect and evaluate.
The company is also opening the door for developers to contribute improvements. Through the public code repository, developers will be able to submit proposed changes for X engineers to review. Not every submission will be accepted, but the company says it is interested in the idea of the algorithm becoming not only visible to the public, but also improved with public input.
However, X is not publishing every part of its internal systems. Some tools will remain private, including certain systems that use Grok to predict whether a post may violate platform rules. X says keeping those systems closed is necessary to prevent spammers and bad actors from reverse-engineering enforcement methods and using that knowledge to bypass platform protections.
The broader goal of this update is to address ongoing questions about how X’s algorithm affects public conversation. Recommendation systems can influence what users see, what topics trend, how political content spreads, and how quickly misinformation can gain traction. Because X remains an important platform for politicians, journalists, creators, businesses, and public figures, its ranking decisions can carry real-world consequences.
Concerns about hidden suppression and algorithmic bias are not new. Even before Elon Musk acquired the platform, users and lawmakers debated whether the company’s systems treated certain political viewpoints unfairly. Claims of “shadowbanning” became a major point of criticism, with some users arguing that their posts were made harder to find without clear notice. The company repeatedly denied that it secretly made users invisible, but the debate highlighted a larger issue: most people had little visibility into how the ranking systems actually worked.
X is now positioning this open source expansion as a way to build more trust. By allowing the public to inspect, critique, and test parts of the ranking engine, the company hopes to show that its systems can be audited and improved over time.
Still, transparency around code does not answer every question about the platform. Since becoming a private company, X has shared less information in some areas than it did when it was publicly traded, including details related to revenue, growth, and certain policy reports. Even so, making more of the For You algorithm available could be one of the company’s most meaningful transparency moves to date.
For users, creators, and developers, the update could make X’s recommendation system feel less like a black box. For researchers, it provides more material to evaluate how large-scale social media ranking systems operate. And for X, it is a chance to argue that its algorithm can be examined, challenged, and improved in public.
Whether this effort will increase trust depends on how much the open source code reveals, how useful the new user transparency tools become, and how willing X is to respond to outside criticism. But the move signals a clear shift: the platform wants more people to look under the hood and understand how the For You feed really works.
