Facebook will let you customize your feed algorithm using AI
Oct 13, 2026, 10:01 AM · The Verge

The Verge’s take on Your Algorithm stresses Meta’s deeper LLM ranking: an algorithm you can talk to, updating recommendations inside the same session.
Why it matters
The Verge reports Facebook’s Your Algorithm lets people add and manage topics for Feed and Reels, with recommendations responding right away after Interested / Not interested updates. Access comes from the three-dot menu on posts, and Meta may ask follow-ups so users can say in their own words what they want more or less of.
Meta says it is using LLMs more in ranking because they can incorporate interests, context, and intent behind natural language—“an algorithm you can actually talk to” that adapts within the same session. The feature previously launched on Instagram, with Threads’ Dear Algo as a sibling control.
From the desk
Same product, sharper framing: Meta isn’t only adding a settings page—it’s wiring language models into the ranking loop so intent becomes a first-class feature. That’s a technical milestone for a feed that spent a decade inferring desire from proxies.
Same-session adaptation is the phrase we’re circling. It promises responsiveness. It also means your offhand “more DIY” can immediately reshape what you see—and what advertisers pay to enter. Power users will learn to steer; everyone else may still swim in the default current.
We’re for interfaces that accept plain speech over hidden knobs. We’re watching whether “talk to the algorithm” becomes meaningful governance or a friendlier skin on the same engagement engine. The test is simple: when users ask for less of something harmful or simply exhausting, does the feed stay changed tomorrow morning?
Context
Meta is standardizing natural-language ranking controls across Instagram, Threads, and now Facebook as LLMs become part of recommendation systems.
Who feels it
- Everyday users
- Lower friction to express feed preferences without learning obscure settings.
- Advertisers
- Faster preference shifts may change auction dynamics and creative targeting.
- Trust and safety teams
- Natural-language intents create new policy edges around disallowed steering.
- Competitors
- Pressure to offer comparable conversational ranking controls.
What to watch
- Session-to-session persistence of user-stated preferences
- Meta’s documentation of how LLM ranking features use conversation text
- Any abuse cases where steering language is used to evade integrity systems
- Cross-app consistency between Instagram, Threads, and Facebook controls
Companies: Meta