Meta is rolling out Your algorithm on Facebook — tell your Feed and Reels what you want in your own words, from the three-dot menu, with recommendations Meta says adapt in real time

2 sources

On October 13, 2026 (Tuesday, calendar-verified), Meta announced: "We're introducing new AI-powered tools that help you shape what you see in Feed and Reels by telling us what you're interested in, in your own words." — "We're rolling out Your algorithm, where you can directly add and manage the topics you want to see in Feed and Reels. Think of it as a dedicated space to tell us exactly what you're into. You can find it in the three-dot menu." The feedback options: "After clicking 'Interested' or 'Not interested,' we may ask follow up questions for more detail on what you liked or disliked." — "You can easily tell us, in your own words, what you want to see more or less of, and your recommendations will update in real time." On the ranking side: "Now, large language models are helping us take a genuine leap forward." — "LLMs can incorporate the interests, context, and intent behind natural language, unlocking an algorithm you can actually talk to." — "Tell it what you want more or less of, and your recommendations respond in real time, adapting within the same session to reflect what matters to you right now." Per TechCrunch (Lauren Forristal, 11:05 AM PDT): "Meta announced the launch of its new AI-powered feature, 'Your Algorithm,' on Tuesday, a feature it first launched on Instagram." — and the outlet's timeline: "Instagram introduced Your Algorithm in June, extending customization options across Feed and Explore after first launching the feature for Reels in December 2025." Threads "also introduced its own related feature, called 'Dear Algo,'".

Why now

The algorithm you talk about on camera just became something your audience can talk to. Per Meta's post, "We're rolling out Your algorithm, where you can directly add and manage the topics you want to see in Feed and Reels." Found in "the three-dot menu", the same menu every Facebook user opens daily. Second layer, the interaction model is genuinely new on Facebook: per the post, "LLMs can incorporate the interests, context, and intent behind natural language, unlocking an algorithm you can actually talk to." The feedback loop starts from the posts themselves, where "After clicking 'Interested' or 'Not interested,' we may ask follow up questions for more detail on what you liked or disliked." Third layer, the behavior claim is Meta's own and worth testing: per the post, "your recommendations respond in real time, adapting within the same session" — a claim your audience can verify by toggling preferences and watching their own Feed. Fourth layer, the context is the outlet's: per TechCrunch, Instagram launched the feature for Reels in December 2025 and extended it to Feed and Explore in June 2026, and Threads has "Dear Algo" — Facebook is, per the outlet, the latest social media platform to give users this control, moving "closer to streaming services that allow users to personalize recommendations based on their interests and preferences".

Why it is worth publishing

The most distribution-relevant card of the day: every creator's audience is on some feed, and the control side of the recommendation loop is now a visible, testable surface. The differentiation play is what nobody should claim: nothing in either source says the feature increases or decreases any creator's reach — it is user-side preference control, and Meta's real-time adaptation is Meta's claim about its own ranking systems. The honest angle is the demo-and-test: walk the three-dot menu, tune a profile live, and show what changes and what doesn't. Creators who publish the honest test own the algorithm-control beat; everyone else will promise 'algorithm hacks', which no source supports.

Evidence basis

Two opened sources: Meta's newsroom post (displayed byline October 13, 2026, Tuesday calendar-verified; article:published_time meta 2026-10-13T16:00:29+00:00 with JSON-LD datePublished matching; opened and read in full) and TechCrunch (Lauren Forristal, byline 11:05 AM PDT October 13, Tuesday calendar-verified; article:published_time meta 2026-10-13T18:05:24+00:00; opened and read in full). Calendar check: October 13, 2026 = Tuesday, matching TechCrunch's 'on Tuesday'. Layer discipline: the Instagram history (December 2025 Reels, June 2026 Feed/Explore) and the Threads 'Dear Algo' sibling exist only in TechCrunch's article — Meta's own post never mentions either — and every history fact rides the outlet's attribution. Meta's post names Facebook Feed and Reels only. Rollout state: 'We're rolling out' — no dates, regions, or language list in either source, never written as available to every user today. Feature-name capitalization stays per-outlet: Meta's title 'Your Algorithm', Meta's body 'Your algorithm', TechCrunch 'Your Algorithm'. The real-time and same-session behavior claims stay Meta's own claims about its own ranking systems. No creator-reach causation is written in either direction — no source supports one.

“Facebook is rolling out a feature where you tell your algorithm, in your own words, what you want more of — and Meta says your recommendations adapt within the same session.”

Angle

Frame it as 'your audience can now talk to the Facebook algorithm — here's the honest test' in three beats. Beat one, the surface: per Meta's post, Your algorithm lives in "the three-dot menu" and lets users "directly add and manage the topics you want to see in Feed and Reels" — show the menu on screen. Beat two, the mechanics: per the post, "After clicking 'Interested' or 'Not interested,' we may ask follow up questions" and "your recommendations will update in real time" — tune a profile live and label Meta's real-time claim as Meta's claim. Beat three, the boundary: per TechCrunch, Instagram got this first (Reels in December 2025, Feed and Explore in June 2026, per the outlet) and Threads has 'Dear Algo' — and nothing in either source says this changes any creator's reach, so say that on camera instead of promising an algorithm hack.

Format

Short talking-head video

Demo idea

A live tuning test: open the three-dot menu on a post, add a topic to Your algorithm, tap 'Interested' with the follow-up question answered in your own words, then scroll the Feed and show what changed — with on-screen labels separating what Meta claims ('respond in real time, adapting within the same session') from what you actually observed.

Platform notes

The rollout state stays 'rolling out' per Meta's post — no dates, regions, or 'available to everyone' claims; if your region doesn't show the menu yet, say that. The Instagram history (Reels December 2025, Feed/Explore June 2026) and Threads' 'Dear Algo' are TechCrunch's facts — attribute them; Meta's own post mentions neither. Meta's post covers Facebook Feed and Reels only. The real-time adaptation is Meta's claim about its own ranking systems — vendor-labeled, never presented as an independent measurement. No creator-reach claims in either direction: nothing in the sources says this feature boosts or starves any creator's distribution. Feature-name spellings stay per-outlet: Meta's title 'Your Algorithm', Meta's body 'Your algorithm', TechCrunch 'Your Algorithm'. zh scripts keep the feature name in English (Your algorithm) and the interface labels in English (Interested / Not interested), and never write 算法被颠覆 or creator-traffic promises.

Usable claims

On October 13, 2026 (Tuesday, calendar-verified), Meta announced new AI-powered tools for shaping what you see on Facebook. Per Meta's newsroom post: "We're introducing new AI-powered tools that help you shape what you see in Feed and Reels by telling us what you're interested in, in your own words." — "We're rolling out Your algorithm, where you can directly add and manage the topics you want to see in Feed and Reels. Think of it as a dedicated space to tell us exactly what you're into. You can find it in the three-dot menu." The feedback options: "After clicking 'Interested' or 'Not interested,' we may ask follow up questions for more detail on what you liked or disliked." — "You can easily tell us, in your own words, what you want to see more or less of, and your recommendations will update in real time." On the ranking side: "Now, large language models are helping us take a genuine leap forward." — "LLMs can incorporate the interests, context, and intent behind natural language, unlocking an algorithm you can actually talk to." — "Tell it what you want more or less of, and your recommendations respond in real time, adapting within the same session to reflect what matters to you right now." Per TechCrunch (Lauren Forristal, 11:05 AM PDT): "Meta announced the launch of its new AI-powered feature, 'Your Algorithm,' on Tuesday, a feature it first launched on Instagram." The outlet's timeline: "Instagram introduced Your Algorithm in June, extending customization options across Feed and Explore after first launching the feature for Reels in December 2025." — and Threads "also introduced its own related feature, called 'Dear Algo,'".

Evidence pipeline

From the news

Breakdown

A recommendation-control story where rollout scope and reach causation are the content. Layer one, the rollout state: "We're rolling out Your algorithm" is Meta's own verb — no dates, no regions, no language list in either source — so the card never writes 'available to everyone today' and never invents a completion date. Layer two, the platform scope is split: Meta's post names Facebook Feed and Reels only, while the Instagram history — "Instagram introduced Your Algorithm in June, extending customization options across Feed and Explore after first launching the feature for Reels in December 2025." — and the Threads 'Dear Algo' sibling exist only in TechCrunch's article and ride the outlet's attribution; the two layers are never welded into one announcement. Layer three, the feature-name capitalization is per-outlet: Meta's title 'Your Algorithm', Meta's body 'Your algorithm', TechCrunch 'Your Algorithm' — attributed mentions mirror the outlet, and the card mints no third spelling. Layer four, the behavior claims stay Meta's: "your recommendations respond in real time, adapting within the same session" and the LLM framing are Meta's statements about its own ranking systems — vendor-labeled, never independent measurements, and the honest creator move is the live test (toggle a preference, watch the Feed). Layer five, the reach boundary: nothing in either source says the feature increases or decreases any creator's distribution — the card writes no causation in either direction, no 'algorithm hacks', no starvation stories. Editor's rules: a rollout with no dates, an outlet-labeled history, vendor-labeled behavior claims, and silence where the sources are silent.

Sources

Risks

Before publishing, re-check each layer: the rollout stays 'rolling out' with no invented dates or regions; the Facebook-only scope holds on Meta's own layer while Instagram/Threads history stays TechCrunch-attributed; no creator-reach causation forms in either direction; Meta's real-time adaptation stays a vendor claim; the feature-name spelling mirrors the outlet being quoted; and the interface labels keep their source punctuation. If your script compresses any of these, cut the detail rather than round it.

Demo ideas

Live tuning test: three-dot menu → add a topic → answer a follow-up question → compare the Feed before and after, with Meta's real-time claim labeled as a claim
Lineage explainer: Instagram's December 2025 Reels start and June 2026 Feed/Explore extension plus Threads' 'Dear Algo' (per TechCrunch), ending with the honest boundary — no source ties this to creator reach