OpenAI shuts down Sora app: AI social network with video failed to find audience

OpenAI is shutting down the Sora app — a mobile client that was positioned as a standalone social feed with videos created by artificial intelligence. Despite the fact that the underlying Sora 2 model continues to demonstrate technically impressive results — one of the best video generators on the market — the AI-social format never found sustained user demand. Sora first came to the attention of a broad audience in February 2024, when OpenAI unveiled its first demonstration videos — realistic footage of cats, surfers, urban landscapes.

The industry's reaction was unequivocal: nobody had ever seen this level of detail, motion coherence, and overall cinematic quality from AI before. In December 2024, the company opened Sora access to a wider audience and simultaneously launched a mobile application with a social feed. A space where you could scroll, like, and share videos entirely created by neural networks — according to the developers' vision, this was supposed to become a new format for social media.

The idea seemed logical: if TikTok turned short video into the dominant content consumption format, why couldn't AI video be the next evolutionary step? The app offered an endless stream of visual content without authors, without subscription algorithms — only pure generation on demand. This concept, apparently, turned out to be the main problem.

Users didn't stick around. According to TechCrunch, interest in the app never became sustained: people would open it, watch a few videos — and leave. TechCrunch called Sora "the creepiest app on your phone" at launch — and not without reason.

Content created purely by AI lacks narrative and personality: there's no author to follow, no story unfolding over time. An algorithm without human curation quickly becomes a monotonous stream of beautiful but meaningless images. Endless videos without context created a feeling of uncanny valley — but no longer in individual frames, but across entire feeds.

There's also a deeper, structural problem with any "purely AI" social network: it lacks a social graph. People return to TikTok and Instagram not just for video content — they return for people. For creators they've come to love, for comments, for a sense of belonging to a community.

The Sora app offered content without creators — and this, judging by the results, turned out to be a dead end. OpenAI decided to redirect resources toward the model itself rather than a standalone application. Sora 2 continues to operate as a video generation tool within the ChatGPT ecosystem and through API.

This is a pragmatic choice: the company gets real data on how professionals — marketers, directors, designers — apply the technology to work tasks, instead of wasting infrastructure on supporting a social platform with an uncertain monetization model. The story of the Sora app is another case in the collection of "good technology doesn't equal a good product." The market knows similar stories: Google Stadia, Amazon Fire Phone.

The highest level of technological sophistication doesn't guarantee product-market fit, especially if the product is built around capabilities for their own sake, rather than around a real user need. Closing the app doesn't mean Sora is a failed project — rather it's a strategic recalibration. OpenAI is betting on B2B integration and professional tools rather than creating its own social network.

The question is who will occupy the AI-video-feed niche: competitors — Runway, Kling, Pika — are actively developing platforms and already testing social elements in their interfaces. The battle for this market is just beginning.