Facebook adds Your algorithm so people can steer Feed and Reels in their own words

What did Facebook launch?

Meta announced on its newsroom on 13 October 2026 a set of what it calls AI-powered tools to help people shape what they see in Facebook’s Feed and Reels. The centrepiece is Your algorithm, “a dedicated space to tell us exactly what you’re into,” where people can directly add and manage topics. It sits in the three-dot menu that appears on every post, which Meta has redesigned so the new controls are easier to reach.

The second part is a feedback flow. People can tap Interested or Not interested on posts in Feed and Reels, and Meta says Facebook “may ask follow up questions for more detail on what you liked or disliked.” Answers can be written in plain language, and Meta says recommendations then update in real time. The company calls it “a more direct, conversational way to shape your experience.”

The pitch is that people should be able to talk to the ranking system rather than only train it through what they tap. As TechCrunch put it, Facebook is the latest social platform to give users more say over the algorithm that decides what appears in their feeds, and the move brings it closer to streaming services that let people tune recommendations to their tastes.

How does Your algorithm work on Facebook?

There are two ways in. The first is the Your algorithm page itself, opened from the three-dot menu. There, people can add topics they want to see more of in Feed and Reels and manage the ones already listed. Meta’s post includes short demo videos but does not say how many topics a person can add, whether there is a cap on how specific they can be, or whether the list starts pre-filled with topics Facebook has already inferred.

The second is feedback on individual posts. The new part is the follow-up. After someone taps one of them, Facebook may ask what exactly they liked or disliked, and the person can reply in their own words, for example asking for more of one sport and less of another. The Verge noted that the buttons may also appear directly under posts, not only inside the menu.

Meta’s announcement does not describe the Instagram features that let people see an AI-written summary of their interests or share them to a Story. On Instagram, The Verge reported in December 2025, users can open an icon of two hearts on slider lines at the top right of Reels to see the topics Instagram thinks they like, remove some and add others. Whether Facebook will get the same summary view is not stated.

What changes inside Facebook’s ranking systems?

The more significant change is the one people cannot see. Meta says its ranking systems have long inferred interests from behaviour, “from the posts you like and comment on to what you watch (among thousands of other signals).” Now, it says, “large language models are helping us take a genuine leap forward.” Because LLMs can read “the interests, context, and intent behind natural language,” Meta describes the result as “an algorithm you can actually talk to.”

In practice, Meta says Facebook brings freeform text together with everything it already knows about a person’s engagement, and that the ranking adapts “within the same session.” That matters because classic recommendation models learn slowly from clicks and watch time; a person who suddenly wants less of a topic often has to scroll past many posts before the feed catches up. A system that can act on a sentence could, in principle, react immediately.

Meta has been adding language models to more of its internal systems. AI Lookout reported last week that it uses an LLM to flag ads that point people to off-platform child-exploitation material, and the company has just launched its own Muse Wonder frontier models. The Facebook post does not say which model powers the ranking feature or whether it runs on Muse, Llama or a smaller in-house model.

How Instagram and Threads got here first

Facebook is the last of Meta’s big feed apps to get this kind of control. Instagram launched Your Algorithm for Reels in the US on 10 December 2025, saying that “using AI, you can now more easily view and personalize the topics that shape your Reels.” At the time Meta also said it would add ways to explain why something in Reels or the Facebook feed was not relevant, and would “introduce new ways for you to shape your Feed” in the following months.

By May 2026, Meta said the feature was available on Reels and Explore in English-speaking countries, and in a 12 May post on teen supervision it let parents see the general topics their teens engage with through Your Algorithm, with notifications planned when a teen adds a new interest. On 10 June, Instagram brought Your Algorithm to the main Feed, with head of Instagram Adam Mosseri saying the company was working on supporting requests for people, moods, content types and more.

Threads took a different route. Its Dear Algo feature, launched in February 2026 in the US, New Zealand, Australia and the UK, lets people write a public post starting “Dear Algo” to change their feed for three days, and others can repost a request to borrow it. Head of Threads Connor Hayes told Fast Company the idea came from users who were already writing such posts.

Why Meta is giving people the controls

Mosseri gave the clearest explanation of Meta’s thinking in June. Algorithmic recommendations are “in many ways a genuine technical achievement,” he wrote, but they have a cost to people’s agency: “The system learns from what you tap, watch, and share, but you don’t really get to tell it what you want.” He argued that this is part of what makes people uneasy about social media, “the sense that the experience is happening to them rather than being shaped by them.”

He also explained why the change is possible now. Ranking models, he said, were built with technologies that are not legible to people, while LLMs can “look at clusters of content and describe them in language people understand.” That gives Meta a way to show people what the system thinks they like, and a way for them to correct it.

There is a business reason too. Mosseri said it is “in our best interest as a business” to let people shape the app. Recommendations that match what people actually want can also keep them using the app longer. Controls over teen feeds also matter as regulators and groups such as Common Sense Media, which recently rated ChatGPT for Teens an unacceptable risk, scrutinise how AI systems treat young users.

What Meta has not said

Meta’s announcement leaves several practical questions open:

• Where and when Your algorithm reaches Facebook users, and whether it launches in English only, as Instagram’s version did.
• How much weight a typed request carries against engagement signals, and how long a request keeps affecting the feed.
• Whether people can see the list of topics Facebook has already inferred about them, as Instagram users can.
• Which language model reads the free text, and whether those typed answers are used for advertising or kept only for ranking.
• Whether parents using Meta’s supervision tools will see the topics their teens add on Facebook, as they can on Instagram.