Intro
For a few weeks last autumn, Sora looked like OpenAI’s next ChatGPT moment: an AI video app racing up the App Store charts, hailed as the future of social media. Now the growth curve has flattened, and the numbers are moving in the wrong direction. That’s more than a blip in one company’s metrics. Sora’s slowdown is an early stress test for the entire idea of “AI-native social networks” — and a warning to founders, investors and regulators about what actually sticks when the novelty of generative video wears off.
In this piece, we’ll unpack what’s really happening behind Sora’s decline, who stands to benefit, and what it signals for the next wave of AI consumer products.
The News in Brief
According to TechCrunch, which cites data from market-intelligence firm Appfigures, OpenAI’s Sora mobile app has seen a sharp drop in momentum only months after its celebrated October launch.
The iOS app, powered by the Sora 2 video-generation model, reportedly surpassed 100,000 installs on its first day despite invite-only access, and quickly hit the number-one spot on the U.S. App Store. It passed 1 million downloads faster than the standalone ChatGPT app. Since then Sora has landed on Android as well.
But Appfigures’ data shows that downloads fell by around 32% month-over-month in December, a period that usually boosts app installs thanks to holiday device sales. In January 2026, installs dropped a further 45%, down to about 1.2 million that month. Consumer in‑app spending has also declined roughly 32% from a December peak of around $540,000 to about $367,000 in January.
Overall, TechCrunch reports Sora has seen roughly 9.6 million installs and $1.4 million in user spending, largely from the U.S., but it has slipped out of the top 100 free apps in the U.S. App Store. OpenAI has not commented on the figures.
Why This Matters
Sora’s early trajectory looked like a textbook case of AI hype efficiently converted into app-store gold. The reversal shows how fragile that formula really is.
First, it exposes the weakness of “wow factor” as a business model. Generating cinematic clips from text prompts is impressive, but most users don’t need to do it every day. Once you’ve flooded your group chats with a few surreal videos of your friends as astronauts or anime characters, the utility curve drops fast. Retention is brutal in consumer apps; AI spectacle doesn’t change that.
Second, it highlights a brutal trade-off between growth and compliance. Sora’s virality partially rode on users remixing famous characters and IP. Once OpenAI tightened copyright controls and moved to an opt‑in model for rights holders — after pushback from Hollywood — a lot of the most shareable content simply vanished. Disney’s later deal to license some characters is too narrow to recreate that chaotic early creativity. Legally necessary guardrails make Sora safer, but also less fun.
Third, it shows OpenAI is not guaranteed consumer dominance. While OpenAI still leads in mindshare, TechCrunch points to fierce competition from Google’s Gemini app and Meta’s AI features in its own social products. Those players don’t need a standalone viral hit; they can inject generative video into feeds users already scroll daily.
The net effect: Sora has become a case study in how hard it is to build a new social graph around an AI capability, especially when rivals can treat that same capability as just one more feature.
The Bigger Picture
Sora’s stall fits a familiar pattern: the AI novelty spike. We’ve seen this movie before with Lensa’s AI portraits, Prisma’s artistic filters, even audio apps like Clubhouse. A compelling new trick drives a huge short-term spike, press coverage and social sharing. Then the question hits: what now?
Several broader trends converge here:
Commoditisation of generative video. When Sora 2 appeared, its output looked almost magical. But the gap is closing quickly. Google, Meta, Runway and others are racing to match or exceed that quality. As models converge, distribution and integration beat raw capability. That favours incumbents with existing social networks, not new standalone apps.
Regulation and IP are shaping product design. The fact Sora had to pivot from opt‑out to opt‑in for IP holders is not just a Hollywood story; it’s a preview of how rights owners will negotiate with all generative platforms. In the past, social apps could “move fast and break things.” With AI and copyrighted characters, legal risk escalates faster than growth.
People are cautious about their likeness in AI. Sora’s distinctive feature — putting yourself and friends into AI-generated scenes — bumps into a real discomfort. Many users do not want their face to be endlessly remixed by acquaintances, let alone strangers. Deepfake scandals and concerns about identity misuse have primed consumers to be wary.
From an industry perspective, Sora’s trajectory echoes Meta’s early metaverse push: impressive demos, unclear everyday use case. We’re learning that “AI-first social network” is not a straightforward product category. The winners may instead be existing platforms that sprinkle AI into proven behaviours: creation tools in TikTok, assistive video editing on Instagram, automated highlight reels on YouTube.
The European / Regional Angle
For European users and companies, Sora’s struggles underline how regulation, culture and business models collide in AI entertainment.
Under GDPR, video content that includes identifiable faces falls squarely into personal data — potentially sensitive biometric data. An app that encourages users to upload their own likeness and those of friends, then enables extensive remixing, runs into strict consent and purpose-limitation requirements. Features that might fly in the U.S. face heavier legal friction in the EU.
The upcoming EU AI Act adds another layer. While generative video tools are not banned, they will likely be subject to obligations around transparency, watermarking and risk management, especially where deepfake-like outputs could mislead viewers. A consumer app blending AI video with a social feed is precisely the kind of product regulators will scrutinise.
Culturally, markets like Germany and the DACH region are notably privacy-conscious. The idea of a quasi‑TikTok where friends and strangers can cast your face into arbitrary scenes is a tougher sell than in the U.S. That creates space for European alternatives that build in stricter consent flows, local moderation and clearer boundaries on identity use.
For European startups in Ljubljana, Berlin, Barcelona or Zagreb that are experimenting with generative media, Sora’s path is a warning and an opportunity: viral growth built on loose IP and ambiguous consent is not a sustainable strategy here. But tools that serve creators, studios, advertisers and game developers — with strong contractual frameworks — may flourish precisely because EU rules demand a more disciplined approach.
Looking Ahead
Where does Sora go from here?
The most likely path is a strategic repositioning away from “AI social network” toward “AI video studio”. Rather than chasing TikTok-style engagement metrics, OpenAI could lean into Sora as an advanced creation layer that plugs into existing platforms: export to Instagram Reels, TikTok, YouTube Shorts, or even professional tools like Premiere.
Expect more licensing deals with rights holders, though probably focused on large entertainment brands rather than the chaotic long tail of memes. Each new IP partnership may bring short spikes in usage, but also raises questions: will users pay for branded models, or stick to free, generic ones from competitors?
On the business side, Sora is small compared with OpenAI’s enterprise and API revenues, but it is strategically important. It’s a testbed for how everyday people interact with generative video, and a showcase for the underlying model. Even if the standalone app plateaus, the technology will likely surface inside ChatGPT, productivity suites, and partner products.
For regulators and civil-society groups, Sora and its peers will be useful case studies in content moderation, deepfake policy and age verification for AI media platforms. Expect guidance and possibly enforcement actions in the next 12–24 months, especially in the EU and UK.
Users and creators should watch for three signals:
- Whether Sora doubles down on pro-creator tools (timelines, editing, export options) rather than social features.
- How aggressively OpenAI localises Sora for non‑US markets, including Europe and Asia.
- Whether Meta, Google or TikTok roll out comparable features at scale — if they do, Sora must either find a niche or risk becoming an impressive demo app.
The Bottom Line
Sora’s slowdown doesn’t mean AI video is a fad; it means viral AI gimmicks are not the same as durable products. OpenAI tried to launch a new social platform on the back of a powerful model, only to run into the hard limits of IP law, privacy fears and shallow use cases. The technology will live on, but likely as infrastructure and tooling, not as a standalone social juggernaut.
The open question for the industry: who will be first to build an AI video experience people actually return to every day — and under what regulatory and ethical constraints will it operate?
