Facebook Updates Reels Algorithm to Prioritize Fresh Content and AI‑Powered Search

Facebook announced a redesign of its Reels feed that pushes newer videos higher in users’ streams, promising roughly 50% more same‑day posts. The change is paired with an AI‑driven search bar that offers suggested queries based on user interests. New friend‑bubble icons let viewers jump straight into direct messages with creators whose Reels they like. AI‑generated videos are treated the same as human‑made content, with the platform relying on likes, comments and “Not Interested” taps to tailor each user’s experience. The updates aim to keep viewers engaged while balancing authenticity and emerging AI media.

Algorithm Shift Toward Fresh Content

Facebook’s product team explained that users have asked to see more recent videos in their Reels feed. Internal testing showed that surfacing posts published on the same day can increase return visits. As a result, the algorithm now prioritizes newer content, delivering roughly 50% more same‑day Reels to each scroll.

AI‑Driven Search Suggestions

A new AI‑powered search tool appears at the top of the Reels interface, offering query suggestions that reflect a user’s demonstrated interests. The feature works similarly to the gray search hints seen on other short‑form platforms, helping users discover videos without typing full searches.

Friend Bubbles and Social Interaction

Facebook introduced small friend‑bubble icons that float in the lower‑left corner of a Reel when a friend has liked it. Tapping the bubble opens a direct message window, giving viewers a quick path to chat about the video. This mirrors a feature previously limited to Instagram Reels.

Handling AI‑Generated Videos

AI‑created content, such as videos generated by tools like OpenAI’s Sora or Midjourney, is treated the same as human‑produced clips in the ranking system. If a user engages positively with AI‑generated material, the algorithm will show more of it; if the user signals disinterest, those videos will be shown less often. The platform also provides labeling options for AI‑generated media, though detection is not perfect.

User Signals Guide the Feed

Facebook emphasized that the system leans heavily on explicit user feedback—likes, comments, shares, and the “Not Interested” button—to refine each person’s feed. Positive interactions boost the visibility of similar content, while negative signals cause the algorithm to suppress that type of video. This approach aims to keep the feed relevant while respecting individual preferences.