Meta Launches 'Vibes' AI Video Feed, Replacing Discover on Its Standalone AI App
Meta has introduced a new feed called Vibes on its Meta AI app, replacing the previous Discover feed with a dedicated surface for short-form AI-generated video content from creators and communities. The rollout extends to the meta.ai website as well, giving the feature distribution across both mobile and browser entry points. The Verge
What Vibes Is and How It Works
Vibes functions as a creator-facing content hub within the Meta AI app. Users can either generate original AI video content directly within the app and publish it to the feed, or cross-post existing AI-generated videos from Meta's broader platform ecosystem — Facebook, Instagram, and adjacent surfaces. Reuters
The mechanics here are straightforward: Vibes is not a separate application but a feed placement inside an app that Meta has been quietly building into a standalone AI product distinct from the social graph. The Discover feed it replaced was a more passive browsing experience; Vibes is explicitly framed around creation and community participation, shifting the surface from consumption to contribution.
The Labeling Layer
Meta's AI content policies for 2025 require that creators label realistic AI-generated images, videos, or audio that could plausibly mislead a viewer. The mechanism is an 'AI Info' label; failure to apply it where required results in content being downranked rather than removed — a calibrated enforcement posture that applies carrot-and-stick incentives without triggering the content-removal debates that have dogged the company for years. Facebook and Instagram users have already been encountering these labels appearing on AI-generated images in their feeds as part of the same broader labeling initiative. AP News
Worth flagging: the downranking penalty is a more measured enforcement tool than outright removal, but it places significant discretionary power with Meta's classifier systems to determine what counts as "realistic" enough to require a label. In a feed explicitly designed for AI-generated content, the line between stylized and misleading is not always obvious, and the practical effect of that determination will fall heavily on creators who cannot anticipate where the threshold sits.
The Advertising and Personalization Connection
Vibes does not exist in isolation from Meta's commercial architecture. In December 2024, Meta announced it would begin using people's interactions with its generative AI tools — including the Meta AI app — to personalize content and advertising across Facebook and Instagram, with that capability going live on December 16, 2025. Reuters
That timeline matters when reading Vibes. A user who engages heavily with AI-generated video content in the Vibes feed is generating a rich behavioral signal — content style preferences, topic affinities, engagement duration — that feeds directly into Meta's personalization and ad-targeting stack once the December 2025 capability is live. The feed is, in that sense, simultaneously a product surface and a data collection instrument, which is a familiar dual function for anything Meta ships inside its app ecosystem.
Competitive Context: OpenAI Is Watching the Same Opportunity
Meta is not the only major AI company moving toward short-form AI video as a consumer distribution layer. OpenAI has launched a social media application built around Sora, its video generation model, explicitly aimed at the short-form video audience currently served by platforms like TikTok. AP News
The convergence is notable. Both Meta and OpenAI are treating AI-generated video not merely as a generative capability to be accessed via API or prompt interface, but as a content format native to social feeds. The implicit bet is that the next major engagement surface on the internet is one where the content itself is machine-generated, and where platform identity accrues to whoever builds the best creation-to-distribution loop.
We have seen this dynamic play out before. In the mid-2000s, when user-generated video began migrating from dedicated platforms to embedded social feeds, the race was not really about the video format itself — it was about which platform controlled the feed algorithm, the creator incentive structure, and the advertising integration. YouTube's eventual dominance came precisely because it got all three right before the incumbents could adapt. The structural question with AI-generated video is whether the feed owner, the model provider, and the ad platform are the same entity — and right now, Meta is the only player in this space that can credibly claim to be all three at once.
What This Means for the Meta AI App's Strategic Position
The Meta AI app has been threading a difficult needle: it needs to be useful enough as a standalone AI assistant to justify user sessions outside of Facebook and Instagram, while also feeding behavioral data back into the core social platforms. Vibes addresses that tension in an interesting way. By making AI video creation and social sharing the primary activity on the feed, Meta gives users a reason to open the standalone app that is rooted in a social behavior — sharing creative output — rather than in utility tasks like querying or summarization.
That approach reflects a reasonable read of where consumer AI adoption is trending. Utility AI — the chatbot that answers questions, drafts emails, summarizes documents — has a high ceiling but a low daily-engagement floor for most users. Creative AI — tools that let people produce something they can show someone else — has historically driven stronger habitual usage, because the social reward loop amplifies the return-to-app incentive.
For technology professionals watching Meta's AI product strategy: Vibes is worth reading as a signal of how Meta intends to differentiate its AI app from the assistant-first posture of competitors. The Discover feed was a browsing surface; Vibes is a participation surface. That shift in metaphor tends to have downstream consequences for engagement metrics, creator acquisition strategy, and ultimately the shape of the ad inventory that gets built on top of it.
The labeling requirements, the December 2025 personalization integration, and the cross-platform cross-posting capability are the three structural pillars holding this up. Each one individually is incremental. Together, they describe a content ecosystem where AI-generated video is the medium, Meta's feed is the distribution channel, and the behavioral data generated in between is the business.
