Facebook Launches ‘Your Algorithm’ AI Tool for Feed Control

Meta is dismantling the opaque recommendation engine that has dictated Facebook content for over a decade, replacing passive engagement tracking with direct, conversational user input.

The company rolled out “Your Algorithm” to Facebook on Tuesday, allowing users to type natural language instructions to dictate what appears in their Feed and Reels. Rather than relying solely on watch time, clicks, or dwell time to infer preferences, the platform now accepts explicit directives. A user can type requests to see fewer political arguments, more local restaurant reviews, or specific hobby content, and the ranking system adjusts those parameters in real time.

This transition fundamentally changes how Meta’s machine learning models process intent. Historically, social media algorithms operated as black boxes, optimizing primarily for engagement metrics that frequently amplified polarizing or sensational material. By introducing natural language processing directly into the curation pipeline, Meta shifts toward a model resembling streaming platforms like Netflix or Spotify, where manual preference inputs carry significant weight alongside behavioral data.

The integration extends beyond a standalone settings page. Meta redesigned the familiar three-dot menu attached to individual posts across Facebook. Alongside existing “Interested” and “Not interested” toggles, users can now access the AI interface directly from any piece of content. The system is designed to be interactive; if a user flags a post they dislike, the AI may prompt follow-up questions to isolate whether the objection targets the creator, the topic, the format, or the tone, refining subsequent recommendations based on those distinctions.

Image Credits:Facebook

Regulatory pressure across multiple jurisdictions accelerated this architectural shift. The European Union’s Digital Services Act mandates that very large online platforms offer users recommendation systems not based on profiling. While U.S. federal legislation remains fragmented, state-level privacy laws and ongoing antitrust scrutiny have pushed major technology companies to build demonstrable user controls into their core products rather than burying them in privacy dashboards.

Facebook represents the final phase of a staggered deployment across Meta’s ecosystem. The company first tested the underlying architecture on Instagram Reels in December 2025 before expanding it to Instagram’s main Feed and Explore tabs in June. Threads received a parallel feature branded “Dear Algo.” Consolidating these tools under a unified framework signals Meta’s intent to standardize algorithmic transparency across its entire portfolio of applications.

Competitors are pursuing similar strategies through different mechanisms. TikTok introduced extensive feed customization filters earlier this year, while YouTube expanded its channel-blocking and topic-muting capabilities. However, Meta’s implementation is the first among major Western social networks to rely on generative AI conversation as the primary mechanism for feed calibration, moving past static toggle switches.

The feature is available immediately to Facebook users globally across both iOS and Android devices.