Meta announced on October 13, 2026 that Facebook will add an AI‑powered feature named Your algorithm that lets users state, in plain language, the topics and types of posts they want to see more or less of in Feed and Reels. The controls are available from the three‑dot menu on posts; changes reportedly take effect in real time and the system may ask follow‑up questions after users mark posts as “Interested” or “Not interested.”
What changed
Official documentation says Facebook now offers a dedicated Your algorithm space where people can add and manage topics that influence recommendations in Feed and Reels. Key points from Meta’s announcement:
- Users can describe preferences using natural language (plain text) rather than selecting only preset interest toggles.
- The three‑dot menu on every post was updated to surface these new controls.
- When users tap “Interested” or “Not interested,” Facebook may follow up with questions so people can explain in their own words what they liked or disliked.
- Meta says recommendations update in real time within the same session after users give textual feedback.
- Large language models (LLMs) are being used alongside Facebook’s existing signals (likes, comments, viewing activity and thousands of other inputs) to interpret intent and context expressed in text and adapt ranking.
Rollout date, status and coverage
Meta posted the announcement on October 13, 2026. The company described the rollout as starting then, but did not publish a complete country list for Facebook in the same post. Reporting from TechCrunch and other tech outlets describes the change as an expansion of functionality Meta previously launched on Instagram and Threads. Sources state availability may vary by account and region; Meta’s post does not confirm immediate global availability for all users.
Who gets it (eligibility and rollout model)
Meta’s documentation indicates the feature is rolling out to Facebook accounts and will be reachable via the three‑dot post menu. Third‑party reports note that similar features (Instagram’s Your Algorithm and Threads’ Dear Algo) initially launched in specific markets before expanding; however, for Facebook the company did not publish a definitive per‑country roll‑out schedule in the announcement. Therefore:
- Eligibility: Available to Facebook accounts as the feature reaches them; no per‑device or account‑type limitations were specified in Meta’s announcement.
- Rollout model: Staged rollout is likely (Meta has used staged rollouts for related features), but Meta has not labelled this release explicitly as a test, limited rollout or generally available. Independent coverage describes it as beginning now rather than fully global.
How to use it (step‑by‑step)
- Open Facebook and find any post in Feed or a Reels item.
- Tap the three‑dot menu on that post; the new Your algorithm option should appear when the rollout reaches your account (Meta updated the three‑dot menu to make the controls easier to find).
- Open Your algorithm to add or manage topics you want to see in Feed and Reels. You can type plain‑language descriptions of interests (for example, “more PC building tutorials” or “less celebrity gossip”).
- Alternatively, tap “Interested” or “Not interested” on a post. If prompted, answer follow‑up questions in your own words to explain what you liked or disliked; Meta says those text responses will update recommendations in real time.
- Observe your Feed and Reels — Meta says recommendations should adapt within the same session to reflect the stated preferences.
Who benefits
- Everyday users who want faster, more precise control over what appears in Feed and Reels without digging through deep settings.
- Creators and small publishers who can better understand how audience preferences might be signaled (for example, clearer requests for specific content types could temporarily influence what a subset of users see).
- Product and content teams who want to experiment with conversational feedback flows as a complement to implicit signals.
Creator and business use cases (realistic examples)
- A tech creator could encourage fans to mark their posts as “Interested” and add short text like “more PC build tutorials” to help the algorithm serve similar content to those fans during that session.
- A local business could ask followers to indicate they want “more event updates” in plain language so those followers see more local event posts in Feed and Reels.
- Publishers might monitor which plain‑language prompts elicit follow‑up questions from Facebook and adjust headline/preview text to match common user phrasing (without gaming the system or using deceptive tactics).
Discovery, ranking and recommendation implications
Meta’s official claim: LLMs help incorporate intent and context from freeform text into ranking alongside existing behavioral signals, and recommendations will respond in real time. That is Meta’s stated design; the company did not provide independent tests, quantitative effect sizes, or guarantees about long‑term ranking changes.
Practical implications (informed inference based on Meta’s documentation and how similar features behaved):
- Real‑time session effects: Expect short‑term adjustments in what you see during the session after giving feedback; Meta states the system adapts within the same session.
- Longer‑term influence: Meta suggests combining text input with accumulated signals will make recommendations more responsive over time, but the announcement does not quantify how persistent changes are.
- No ranking guarantees: Meta did not promise increased reach, monetization, or ranking benefits to creators; do not interpret user controls as a method to boost algorithmic reach beyond normal engagement‑driven signals.
Privacy, safety and limitations
- Privacy: Meta’s post indicates the text you type will be used by ranking systems together with existing signals; the announcement does not provide granular data‑retention or model‑training details in that post. Users concerned about how text feedback is stored or used should consult Meta’s full privacy documentation and product support pages for specifics.
- Abuse and manipulation risks: Meta did not publish defenses specific to the plain‑language input in the announcement. As with any user‑provided content, there is potential for misuse; Meta’s broader content and ranking policies remain applicable.
- Adoption friction: Past research and reporting show most users rarely change settings; Meta itself acknowledged that users often do not manually refine settings. The feature’s effectiveness depends on adoption and clarity of prompts.
- Unknowns: Meta did not disclose whether the feature uses user text to further train foundational models, how long session adjustments persist, or the full list of countries where the controls are live for Facebook at launch.
Practical checklist for creators and social teams
- Verify availability: Check the three‑dot menu on Feed/Reels posts to confirm Your algorithm appears for your account.
- Encourage authentic signals: Ask engaged followers to use “Interested” and describe what they want in plain language if they want to shape their personal feed—avoid requests that try to game ranking or ask for fake engagement.
- Match language: Use the same natural phrasing your audience uses in headlines, captions and CTAs so user input and your content language align organically.
- Monitor analytics: Compare short‑term changes in impressions/engagement after followers use the controls; don’t treat this as a substitute for consistent quality and retention tactics (hooks, watch time, retention, captions, community interaction).
- Review privacy docs: Update internal guidance on how user feedback might be processed and inform community managers about any moderation or policy implications.
Evidence comparison and context
Meta’s newsroom post is the primary source for feature mechanics and claims. Independent tech reporting (TechCrunch, Social Media Today, Digital Trends, Unite.AI and others) uniformly describes the feature as an extension of Instagram’s and Threads’ earlier controls, and they report that the controls respond in real time and are accessed via the three‑dot menu. Where sources differ or omit detail, note:
- Scope and timing: Meta’s post gives the launch date (October 13, 2026) but does not publish a full country rollout map for Facebook; some secondary reports state availability may vary and reference staged rollouts used on Instagram and Threads.
- Persistence and training: Meta describes real‑time session updates but did not quantify how persistent the changes are or whether user text will be used to further train LLMs; secondary outlets reiterate Meta’s claims without new technical confirmation.
What remains unknown
- Full per‑country availability and precise eligibility criteria for Facebook at launch.
- Whether plain‑language feedback is stored or used for additional model training and for what retention period (Meta’s broader privacy pages are the next reference point).
- Quantitative impact on impressions, distribution or long‑term ranking behavior; Meta provided no numbers in the announcement.
Bottom line for U.S. users and social teams
Facebook’s Your algorithm brings conversational, plain‑language controls into Feed and Reels and is designed to let people shape recommendations faster than by relying on implicit signals alone. The feature is rolling out as of October 13, 2026; availability appears staged and not yet documented by Meta on a per‑country basis for Facebook. Creators and community managers should treat the controls as a user‑side tool to improve relevance for individual accounts, not as a guaranteed growth lever—use it to encourage authentic feedback, align content language to audience phrasing, and monitor analytics for short‑term shifts.