Meta explains Facebook Your Algorithm controls

Meta has published an engineering explainer detailing how Facebook’s Your Algorithm controls translate user preferences into changes to content recommendations.

The tool lets users add or remove topics from an automatically generated list of interests based on their in-app activity. A recently added chatbot prompt also lets users describe what they want to see more of.

How Facebook weights topic preferences

According to Meta, newly added interests receive extra weight for a limited period so users can see the change. After that, those interests are weighted according to user interaction, like other topics.

The company said the system also balances exposure across topics to avoid overwhelming users with any particular subject.

Large language models generate topic profiles and check user-made changes to ensure they fit into a topic category. Meta explained: “An LLM turns interest signals, which are refreshed regularly, into a set of inferred interests with clear, readable topic names.”

How requests for less content work

Meta distinguishes topic preferences from broader requests about content quality.

“Quality-related requests, such as reducing AI-generated or sexually suggestive content, map to dedicated demotion signals rather than topic matching,” the company said.

During ranking, the system removes content candidates that are too close to an interest marked “Not Interested In,” while demotion rules push down near matches, according to Meta.

The original report noted that these controls can reduce AI-generated content in a Facebook feed without eliminating it entirely, and that future engagement will still influence ranking.

Requests to see less from a specific creator work differently: Meta said “that creator’s content is filtered out directly.”

The source report, published on November 11, 2026, also described a policy development: last month, the Australian government began preliminary investigations into potential laws that would require social media platforms to offer users an option to switch off algorithmic recommendations and make non-algorithmic feeds the default.