There is no place for half measures on the artificial intelligence market. When OpenAI in 2024 showed the world Sora – AI video generator, the reactions were mixed: delight over the possibilities and … disappointment over imperfections.
On September 30, the premiere of Sora 2 took place, and with it a new social application in the Tiktoka style. The AI industry has just entered the next lap of the race, in which the stake is not only innovation, but also our habits in using the network and social media.
Sora 2: “GPT-3.5 Moment” for video?
On September 30, Opeli announced the premiere of Sora 2 – continuation, the evolution of the Sora, which is not so much a blast of mistakes, but puts a new standard in generating movies.
The team responsible for the project says directly:
We jump to the moment that can be for video what GPT-3.5 was for the text
These are strong words, considering the fact that the competition has been done in recent months. Sora 2 can generate clips that visually do not differ from a material recorded with a camera. Physics models? Are. Adding audio and lipdub? Is.
- Gymnastic jumps at the Olympic level – with the realism of movement and dynamics of the body.
- Salta on the SUP board – With the physics of displacement and stiffness.
- Triple Axel … with a cat holding at all costs – It sounds like a meme, it looks like a fragment of professional assembly.
The greatest allegation against AI Video, i.e. characteristic hallucinations and coarse assembly, begins to disappear slowly. It is worth emphasizing, however, that Sora 2 transports what is already available on the market using VEO3, Hailuo Minimax or Kling AI.
Sora App – Tiktok, but fully powered AI
Together with the new Openai generator, he introduces on iOS Sora App, i.e. a social application that acts like a hybrid of the thicket and a creative laboratory. Difference? You won’t upload your own movies or photos there. Everything you watch and share was created thanks to AI.
- Length limit: 10 seconds.
- Feed: algorithmic, but the user has more control than in classic social stories.
- The most interesting function: Cameos – After confirming your identity, the application creates your digital alter ego, which friends can use in their films (with your consent). You get a notification, even if the movie has not been published. Deepfake of the new era.
Tester feedback? Cameos to Game-Changer. Like simple, but creates a completely new way of interaction – instead of commenting, you can … put yourself into someone’s scenes. On the one hand, it gives enormous possibilities, but on the other, it poses huge challenges regarding safety and inadequate use of such possibilities of Sora App.
Opeli focuses on creation, not scrolling
At a time when Tiktok and Instagram race on a screen time, OpenAi plays differently.
The company declares that it does not optimize the application in terms of time spent in Feed. The goal is to be creation, not consumption.
For teenagers – daily content limits and Cameo restrictions.
For parents – Integration with new parental controls in Chatgpt.
For everyone – No advertising and free access to start (although in the future there will be fees for generating additional films during rush hour).
First reactions: shock and creative explosion
Although the application and the generator itself remain invite-only, the internet has already boiled from experiments. One of the users generated a film in which Altman himself steals work from the Ghibli studio, and Hayao Miyazaki is trying to recover it – a subtle reference to the spring scandal from GPT -4O. Effect? So refined that even audio is difficult to distinguish from a real recording.
A profession or a breakthrough?
The question is whether Sora 2 is another profession or actual conclusions, can be summarized briefly: this time OpenAi made it. Of course, barriers still exist – limiting the length of films, lack of full control over the narrative or the closed nature of the application. But if previous generators were an early sketch, then Sora 2 is almost a full -fledged storyboard.
Is this the moment when the video generated by AI will become as common as photos of Midjourney on Instagram? Probably yes. But this is also the moment when the border between creation and manipulation begins to blur dangerously. Deepfake will now be able to do everyone at a minimum cost and effort.
