Openai officially took part in the short video war. On Tuesday, the company launched Sora, a new Invite-only iOS app that can generate AI-powered videos using text or images.
Built on the new SORA 2.0 model, the app is still pitched as Openai's most advanced video tool and comes straight on platforms like YouTube and Tiktok, but it involves camera and editing timelines.
SORA 2.0 is expanded with previous versions dropped this year as a research preview. The upgrade now supports multi-shot sequences, hyper-real visuals and synchronous audio from all single prompts.
I'm running a new app called Sora. This is a combination of a new model called Sora 2 and a new product that makes it easy to create, share and view videos.
This feels like a “chat for creativity” moment for many of us, and it feels fun and new. There's something…
– Sam Altman (@sama) September 30, 2025
The app allows users to write scenes, drop images, and create cameos with videos. Openai promises to add strict controls for portraiture, safety and source. Users get tools to remix scenes and insert themselves, but only if they see their identity. If you don't sign up for a cameo, Openai says it won't show up.
The studio warned as copyright concerns grew
The content moderation policy around Sora has already created noise. According to CEO Sam Altman, Openai begins sending notifications to studios and talent agencies, warning that unless they explicitly opt out, their copyrighted material may appear in content generated through SORA.
Openai claims this is no different from how you handle things with image tools. The company says it considers this to be a “fan expression.” Here, users are remixing or referring to their already beloved fictional world.
Executives said early testers on the platform were clearly interested in interacting with existing characters, building content around known franchises, and creating their own versions of familiar scenes.
Still, Openai depicts a hardline between fictional characters and real people. Users may be able to recreate Marvel-style universes or create Harry Potter imitations, but they cannot generate a person's face unless the person gives written approval.
No, eating someone's image to create a deepfake doesn't work. That's also blocked. anyway.
Openai expands Sora to audio and Hollywood
Sora doesn't just make videos. It also uses AI to create background sounds, dialogs, and sound effects, perfectly timed with the video. And yes, it supports multiple languages. This allows creators to easily publish AI movies, trailers, or ads without studios or voice actors.
These features are already attracting Hollywood's attention, despite being filled with the result of a mixed result of Openai's efforts to plead the industry.
During the demonstration, Openai gave two very different examples. One was a fake news report showing a man named Dimeson drinking ketchup from a MacDonald's dispenser.
The other was an AI-maid perfume ad for “Sora 2.” “A new scent from Sora – fresh, clean, non-apologetic, for the person you choose.”
The app is currently only available in the US and Canada, with only invitations access. The pro version, web access, and developer API are already working. But the company hasn't put this tool in hand.
All videos made with SORA are watermarked and exports are marked with those marks too. Screen recording is disabled by default to stop people from copying and reposing videos without labels.
This tool will not allow public figure videos unless those individuals validate cameos and opt in. This applies to both video prompts and image input. If you are famous and didn't ask to appear, you won't. According to the company, all control layers have been built to reduce misuse.
Join Premium Crypto Trading Community for free for 30 days – usually $100/month.
