The move is a bid by Facebook's parent company to address the threat of TikTok, reversing its original plans to show more family and friends' post.
Meta Platforms Inc said on Thursday it was revamping the main feed on its Facebook app to prioritise "discovery" of new content, instead of posts from accounts users follow, a bid to style its apps after short-form video competitor TikTok.
Meta executives have voiced increased urgency in recent months around boosting the company's "Reels" product, similar to TikTok's short video format that has attracted many younger users.
"Home", Facebook's main news feed tab that users will see when they open the app, will start more heavily featuring popular posts from accounts that users do not follow, including Reels and Stories, Meta said in a statement.
Facebook will suggest posts to users with its machine learning ranking system and is investing in artificial intelligence (AI) to serve recommended content, it added. A new separate tab called 'Feeds' will offer a version of the old approach, which overwhelmingly features posts from friends, pages and groups that users actively choose to follow.
Threat from TikTok
Feeds in that tab will be presented chronologically, without personalised ranking, Chief Executive Officer Mark Zuckerberg said in a Facebook post. Meta said the feeds will not include suggested posts but will still have advertisements.
The world's biggest social media company has gone all in on algorithmic recommendations in recent months as the threat from TikTok has grown, a stark change from its 2018 plan to feature more posts from friends and family in the news feed.
Its Instagram app announced tests of a more "immersive" TikTok-style viewing experience in May, while Zuckerberg told investors in April that Meta was making significant investments to support the "discovery engine" approach.
Earlier this month, Chief Product Officer Chris Cox told employees there was a plan to increase fivefold the number of graphic processing units (GPUs) in its data centers by the year-end to provide extra computing power for AI.