Facebook Lets Users Steer Feed Recommendations With Natural-Language AI Controls

Facebook is rolling out Your Algorithm, an AI-powered feature that lets people directly shape the topics appearing in Feed and Reels. Users can describe what they want to see in ordinary language, then mark subjects as interested or not interested to adjust recommendations. Meta announced the feature on October 13, 2026, after introducing a similar control on Instagram.

The feature gives users a more direct role in a system that has traditionally inferred preferences from clicks, views, likes, follows and other behavior. The Verge AI says recommendations respond right away after users update their preferences. TechCrunch AI describes the tool as a way to influence both connected posts and recommended content, while Meta’s broader explanation of its ranking systems says multiple machine-learning models select and order material in the feed.

Your Algorithm translates stated preferences into signals for Facebook’s recommendation systems. A user might ask for more posts about a subject or indicate disinterest in another topic, then see the feed adjust without waiting for the system to infer that preference from repeated behavior. The controls apply to Feed and Facebook Reels.

Facebook still combines direct feedback with its existing ranking process. Connected content includes posts from friends, groups and Pages a person follows, while recommended content comes from accounts and topics outside that existing network. Socialbee and Socialpilot describe the feed as a prediction engine that ranks candidates according to expected relevance and engagement, with discovery playing a larger role than the older follow-only model.

The rollout addresses a long-running complaint about opaque social-media ranking. Facebook has faced scrutiny over how its systems determine which posts receive attention, and earlier controls included Favorites, Snooze and settings that changed the balance between friends, groups and Pages. A 2021 Facebook announcement, reproduced by BBC, described tests that let people increase posts from friends and family while reducing posts from groups and Pages.

Your Algorithm shifts that idea from buried settings toward conversational preference management. Meta’s 2023 explanation framed personalization as a process shaped by both automated ranking and user choices, while the new feature makes those choices more explicit. That distinction matters because users gain a visible feedback loop: they state a preference, see recommendations change, and refine the result rather than simply accepting an opaque stream.

Creators and businesses still compete for attention inside an AI-ranked feed, but audience discovery becomes more dependent on declared interests as well as engagement history. A page that publishes content matching topics users actively request has a clearer route into recommendations, while irrelevant or repetitive material faces more direct rejection.

The feature doesn't remove Facebook’s ranking models or guarantee distribution. Socialpilot and Quso emphasize that consistency, meaningful engagement, video and content quality remain central to reach. Their guidance reflects the continuing importance of audience response, even as Facebook adds a stronger user-controlled layer. Businesses will need to understand which subjects their audiences actively select, not just which formats generate passive impressions.

Your Algorithm gives users more influence, but Facebook retains control over the recommendation infrastructure, eligible content and ranking signals. Meta’s feed documentation says several machine-learning models work together and that their inputs change as the system learns and is updated. A stated preference therefore guides ranking without replacing the broader system.

The quality of the experience will depend on how accurately Facebook interprets natural-language requests and how consistently those instructions affect recommendations. Users also still receive a mix of posts from accounts they follow and material selected for discovery. That balance preserves Facebook’s recommendation business while giving people a clearer way to correct unwanted themes.

Facebook’s launch extends a broader shift toward user-steerable recommendation systems. Instagram already introduced Your Algorithm, and Facebook is now applying the model to a larger network that mixes personal updates, communities, Pages and short-form video. The change gives Meta a way to answer demands for greater control without abandoning personalized ranking.

For users, the practical test is whether direct instructions produce visible, lasting improvements. For creators, the test is whether topic relevance translates into reliable discovery. Facebook’s feed will remain an AI-ranked system, but the interface now gives people a more active hand on the steering wheel.