Meta Lets Facebook Users Rewrite Their Feed With AI

Meta is giving Facebook users a new kind of control over what they see. The company is rolling out a set of AI-driven customization tools that let people describe, in ordinary language, the content they want in their feed and in Reels, the short-video format that competes with TikTok. The feature is called Your Algorithm, and it is the clearest sign yet that the company wants users to feel they can steer the machine rather than merely react to it.

The mechanics are simple. A three-dot menu on any post now offers a new option. Tap it and you can mark a topic as interesting or not interesting. The system then asks follow-up questions to sharpen your preference, and a large language model translates your answers into instructions for the ranking algorithm. Instead of guessing from clicks alone, Facebook is asking directly, and in natural language, what you want more of.

For years, social platforms treated recommendation as a black box. Users could hide a single post or unfollow a friend, but they could not easily reshape the broader mix of content the algorithm served them. That opacity drew scrutiny from regulators and researchers who argued that engagement-optimized feeds can amplify outrage and misinformation. Meta’s new tool is partly a response to that pressure, and partly a product bet: if people can tune their feed, they may spend more time on the platform instead of drifting away.

The timing matters. Meta has been spending heavily on artificial intelligence, from data centers to model development, as it tries to keep pace with rivals in the AI race. Features like Your Algorithm turn that investment into something ordinary users can touch. They also generate valuable signals. Every time someone tells the system what they like, Meta collects a cleaner, more explicit preference than a mere pause or a like, which can be noisy and ambiguous.

Rivals are watching closely. TikTok has long built its reputation on a powerful recommendation engine, and YouTube has experimented with user controls over recommendations. If Meta’s natural-language approach works, it could become a standard way for platforms to let people negotiate with algorithms. That would mark a shift from passive consumption toward something closer to a conversation with the software that decides what you read, watch and share.

There are limits. Users must still trust that the system honors their instructions, and early controls on social platforms have often been partial. Marking a topic as not interesting does not guarantee it disappears, and the algorithm still balances personal preferences against broader engagement goals. Nor is it clear how much of the feed will truly bend to user wishes versus nudging them toward content that keeps sessions long.

Even so, the direction is notable. For a global audience, the story is about who controls attention online. Facebook, with billions of users, is testing whether people want to program their own feeds the way they might program a playlist. If the answer is yes, the humble three-dot menu could become one of the most consequential buttons in social media, quietly shifting power from the algorithm back toward the person holding the phone.