Meta has launched “Your Algorithm” on Facebook. It’s an AI-powered feature that lets you shape your Feed and Reels by describing, in your own words, what you want to see more or less of. TechCrunch AI reports that Meta announced the launch on Tuesday, after first rolling the feature out on Instagram. So instead of guessing what you like from every scroll and pause, Facebook will now let you tell it.
This matters because recommendation engines have mostly been black boxes. You could hide a post or unfollow a page, and that was about it. Now Meta is giving you a conversation with the system that decides what fills your screen.
What Facebook Actually Shipped
Here are the main pieces, based on TechCrunch AI’s reporting:
- Preferences in plain language. You don’t pick from preset categories. You type what you’re into, or what you’re tired of, the way you’d explain it to a friend, and Facebook adjusts your Feed and Reels to match.
- Changes happen in real time. Meta says your adjustments show up in your recommendations right away. There’s no waiting days for the algorithm to slowly catch on, and that quick feedback is what makes the tool feel responsive.
- A reworked three-dot menu. The menu on each post now gives you direct access to “Your Algorithm,” plus “Interested” and “Not interested” options. You can flag a post while you’re looking at it, so you don’t have to dig through settings.
- Follow-up questions. Facebook may ask what exactly you liked or disliked about a piece of content. A thumbs-down on one cooking video might mean you hate the creator, the recipe, or the format, and the follow-ups help the system tell those apart.
- Ongoing refinement. You can keep explaining what you want more or less of, and Facebook will keep tuning its recommendations based on what you say.
How It Fits Meta’s Bigger Plan
Facebook isn’t first here. It’s catching up with its sister apps.
- Instagram first launched Your Algorithm for Reels in December 2025, then expanded it to Feed and Explore in June.
- Threads has its own version, called “Dear Algo,” which also lets users steer what gets recommended to them.
With Facebook joining, Meta now has natural-language feed controls across its three biggest social platforms. Expect the company to keep using what it learns from one app to improve the others.
Why This Is a Real Shift
What stands out here is the change in who’s driving. For years, social feeds ran on passive signals like watch time, likes, shares and how long you lingered. You never told the algorithm what you wanted. It worked that out from how you behaved, and it often got things wrong in annoying ways.
TechCrunch AI points out that this brings Facebook closer to streaming services, where you shape your recommendations around stated interests. It’s also a natural use for large language models. Turning a sentence like “less political drama, more woodworking tutorials” into ranking signals is exactly the kind of fuzzy interpretation these models handle well.
There’s a competitive angle too. Other platforms have been adding similar controls, and users have spent years complaining that feeds feel random, manipulative or stuck on stuff they don’t care about. Giving people a visible way to push back is good for Meta’s image as well as its engagement numbers.
Caveats Worth Noting
The announcement leaves some open questions:
- Influence, not full control. You can steer the algorithm, but Meta still runs it. Your preferences are one input among many, and ads and other ranking factors aren’t going away.
- More data about you. Every preference you type is a detailed, explicit signal about your interests. That’s useful for your feed, and just as useful for Meta’s ad targeting.
- Rollout details are thin. The report doesn’t say whether every region gets it at once or whether some users will see it later. The feature lives inside the standard Facebook app, and there’s no sign of a paid tier.
What Comes Next
Feeds you can talk to look set to become the norm. Once people get used to telling an algorithm what they want, the old “just trust the machine” approach will look outdated. The real test is whether Facebook’s recommendations actually follow what users say, or whether the familiar engagement-driven content slowly creeps back in. Full details on the launch are available at TechCrunch AI.