Meta launches AI search mode in Facebook, promising local recommendations

Meta is betting that Facebook can answer not just where to go, but what to do, by turning public posts into AI-generated recommendations. The problem is trust: when a system pulls from Groups, Reels and other public chatter, rumor, sarcasm, spam and hyperlocal gossip can all look like authority if the answer arrives polished and confident.

Meta launched AI Mode on Facebook on June 15, 2026, as part of a broader AI update. The company described it as a search tab powered by Muse Spark that uses Meta AI to return answers rooted in the culture, opinions and recommendations people share publicly across its apps, rather than simply listing links. Meta said the feature is meant to surface “real perspectives and experiences,” positioning Facebook search as a kind of recommendation engine for local advice and weekend planning.

That promise comes with obvious risk. A response based on public posts may be current, but it may also be incomplete, contradictory or flat-out wrong, especially when it is stitched together from user-generated content that was never written as a fact-checked source. The visibility problem is just as serious: if an answer draws from a mix of posts, it is not always clear to the user where the recommendation came from or why one post was favored over another. For a platform with billions of posts and rapid-fire local conversation, source transparency is not a cosmetic issue. It is the difference between a helpful shortcut and a misleading summary.

The launch lands against a difficult record for Meta’s AI efforts. In July 2024, the company acknowledged that Meta AI gave incorrect responses about the Trump rally shooting. Meta also revised its AI-content labeling after finding the labels were not always aligned with people’s expectations and did not always provide enough context. Those missteps matter more now because Meta has also moved away from third-party fact-checking in the United States and toward a Community Notes-style moderation system, a shift that revived misinformation concerns in 2025 just as the company pushed AI deeper into Facebook.

Meta said camera roll sharing suggestions in the new Facebook AI rollout are opt-in and can be turned off at any time, a sign that the company knows the line between convenience and creepiness is thin. But the bigger test for AI Mode is whether it can distinguish a useful neighborhood tip from an online hallucination of consensus. If it cannot, Facebook may become faster at answering local questions while becoming less trustworthy at the same time.