Facebook Now Lets You Steer Your Feed With Plain Words

Meta has launched its AI-powered "Your Algorithm" feature on Facebook. The company announced the rollout on Oct. 13 under the title "Your Algorithm, Your Words: New Ways to Shape What You See on Facebook" Meta. The feature came to Facebook after first appearing on Instagram TechCrunch.

On Facebook, "Your Algorithm" shapes Feed and Reels. It lets users state preferences in natural language, meaning plain written instructions rather than code TechCrunch. The input is free-form text rather than a settings toggle. A user can ask for more of a subject or less of it, and recommendations adjust accordingly.

Those changes appear in recommendations in real time TechCrunch. A new instruction changes what shows next without a waiting period. Access is built into existing menus. Facebook updated its three-dot menu on individual posts to manage content preferences at the post TechCrunch. That menu now lists "Your Algorithm" alongside "Interested" and "Not interested."

Instagram added "Your Algorithm" in June for Feed and Explore, after first offering it for Reels in December 2025 TechCrunch. Threads added a related tool called "Dear Algo" to shape recommended content. Meta's Transparency Center describes both tools as ways to shape what is seen by typing in preferred interest topics Meta Transparency Center.

The broader context here is feed ranking, the system that orders posts. Until now it relied largely on inferred preference, or guesses from clicks and watch time. This adds stated preference, or direct instructions. It is like moving from fixed buttons to a dial that can be set precisely. "Interested" or "Not interested" can hide one post. A sentence can name a topic, a tone, or a boundary.

In my view, the test will be persistence and composition. A text control must sort out clashes between older instructions and current clicks, and between lasting interests and passing curiosity. Fast updates help. Usefulness will rest on how well the system remembers earlier directions and lets users revise them without starting over.

The wider point here goes beyond social feeds. If users take to tuning feeds with words, the same method could help enterprise search, internal knowledge ranking, and consumer discovery, which all deal with personalization that is hard to inspect. Giving users plain words to steer the system is a modest step that could make those tools easier to use.