OpenAI just dropped what could be the most controversial social app of 2025. The company's new TikTok-style Sora app for iPhone lets users create AI deepfakes of their friends with consent, powered by the upgraded Sora 2 video model. Company executives are calling it the potential "ChatGPT moment for video generation" as the invite-only platform begins rolling out across the US and Canada.
OpenAI is betting big that people want to deepfake their friends - with permission, of course. The company's new Sora app for iPhone transforms the controversial AI video technology into a social media platform that looks and feels like TikTok, complete with scrollable feeds and remix features. But instead of dance challenges, users are creating AI-generated "cameos" of each other using the upgraded Sora 2 model.
The timing couldn't be more strategic. As TikTok faces ongoing regulatory pressure and Meta struggles to capture younger audiences, OpenAI is positioning itself as the next evolution of social video. "We think this could be the ChatGPT moment for video generation," company employees told reporters during a Monday briefing, drawing parallels to their chatbot's explosive 2022 launch.
The consent mechanism represents OpenAI's attempt to solve deepfake ethics before they become a problem. Users must explicitly upload a "cameo" video of themselves before anyone can generate AI content featuring their likeness. The person being deepfaked becomes a "co-owner" of any generated content and can delete videos or revoke access at any time, according to OpenAI's official release.
But the app's limitations reveal just how carefully OpenAI is treading. Videos max out at 10 seconds, public figures can't be generated without consent, and the company claims it's "impossible" to create adult or extreme content. Currently, only US and Canada users can access the invite-only platform, with each user receiving four additional invites to share.
The technical leap from Sora's December 2024 launch to today's social app showcases rapid iteration. Sora 2's ability to maintain facial consistency while generating contextual video content represents a significant advancement in AI video synthesis, potentially leapfrogging competitors like Google's Veo or Meta's Make-A-Video.
The social implications extend far beyond entertainment. If Sora gains traction similar to ChatGPT's viral adoption curve, it could fundamentally alter how people create and consume video content. Traditional social media platforms built around user-generated content might find themselves competing against AI-generated alternatives that require minimal creative skill or equipment.
