Meta has decided to roll out yet another experiment in AI-driven content: a new feature called “Vibes.” The update is debuting inside the Meta AI app and on meta.ai, offering a feed where users can share and create short-form, AI-generated videos. In simple terms, it’s Meta’s answer to TikTok or Instagram Reels—but with a twist: instead of human creativity, the feed is filled entirely with algorithm-made clips.
CEO Mark Zuckerberg announced the launch on Instagram, showcasing examples of the new format. Among the previews were a group of fuzzy digital creatures hopping across cubes, a cat casually kneading dough, and even an ancient Egyptian woman snapping a selfie with pyramids in the background.
The Vibes feed allows users to either generate videos from scratch or remix existing ones they encounter. Before publishing, creators can add music, layer visuals, tweak styles, and then share directly within Vibes. Content can also be cross-posted to Instagram and Facebook Stories and Reels or sent via DMs. Meta says the algorithm will adapt to user preferences over time, curating a personalized AI content stream.
To power the early version of Vibes, Meta has teamed up with well-known AI players like Midjourney and Black Forest Labs, though the company is also refining its own in-house models.
Despite the polished pitch, early user reactions weren’t flattering. On Zuckerberg’s announcement post, top comments included: “Nobody wants this,” and “Bro’s posting AI slop on his own app.” Many expressed confusion and fatigue over the flood of AI-generated media already swarming social platforms.
The launch comes at a curious moment. Earlier this year, Meta stressed that creators should focus on “authentic storytelling” instead of low-value, recycled content. Now, the company appears to be fueling the very trend it warned against.
This release also reflects Meta’s broader push to reassert itself in the AI race. Following concerns it was lagging behind competitors like OpenAI and Google DeepMind, Meta restructured its AI division into the newly branded Meta Superintelligence Labs. The company split its teams into four groups, focusing on foundation models, research, product integration, and infrastructure—signaling just how high the stakes are in the AI arena.
