OpenAI Releases Sora AI Video Social App, Challenges TikTok and Meta
- tech360.tv
- Oct 2
- 3 min read
OpenAI has launched a standalone social application for generating and sharing artificial intelligence-powered videos, marking its largest stride yet towards building a social media product. The free Sora app aims to supercharge adoption for the emerging technology, much like ChatGPT did for chatbots three years ago.
Available by invitation from Tuesday, the Sora app is powered by a new version of OpenAI’s video-making software of the same name. Users can create short clips from text prompts and also view videos made by others.
This move places OpenAI in closer competition with services from TikTok and Meta Platforms, which recently introduced an AI video feed named Vibes. The new app could also open doors to fresh revenue streams from advertising and boost the visibility of OpenAI’s technology.
Users can create a realistic-looking AI avatar and voice of themselves, which can be inserted into videos made with the app by the user or their friends, with the avatar owner’s permission. This feature allows users or their friends to include the avatar in generated content.
Thomas Dimson, a software engineer at OpenAI, shared insights during a briefing this week. Dimson mentioned internal scepticism about an AI-generated feed, but the cameo feature, allowing users to insert themselves into videos, shifted perceptions, with staff believing it could be “good for the world.”
Dimson noted, “Over time, I’ve noticed that a lot of the other social platforms have actually drifted away from this idea of connections and friends.”
The updated Sora software is designed to improve longstanding flaws in AI video generators. OpenAI stated the new software is better at generating videos whose visuals adhere to the laws of physics, a common challenge for previous services, including the original Sora version.
It also aims to follow user prompts more reliably, especially across a sequence of different shots within a video. Bill Peebles, who leads OpenAI’s Sora team, expressed optimism, stating, “We really think we have something which feels like it might be the ChatGPT moment for video generation.”
Peebles highlighted the software’s ability to create more complicated visuals, such as a person doing a backflip atop a paddle-board in a body of water, complete with proper fluid dynamics and buoyancy. It can also automatically generate and assemble individual scenes.
The new Sora can generate and synchronise audio, including sound effects, background noise, and dialogue in multiple languages. These features could make Sora a bigger draw in Hollywood, an industry OpenAI has been courting for months with mixed success.
During a press demonstration, OpenAI showcased a news-style clip of Dimson drinking ketchup from McDonald’s dispensers. Another AI-generated video resembled a perfume advertisement for Sora 2, featuring the tagline: “The new fragrance from Sora – fresh, clean, unapologetic, for whoever you choose to be.”
As Sora’s capabilities improve, concerns may arise about its impact on the filmmaking industry and the distinction between authentic and AI-generated content online. OpenAI stated it would not permit users to create videos featuring public figures, unless those individuals choose to make a cameo of themselves.
The organisation also confirmed it would not allow the generation of videos of people based on an image. Videos created with the Sora app will be watermarked to indicate their AI origin.
The same watermarking will apply to videos exported from the app. OpenAI is also disabling screen recording for the app to help manage video sharing.
The Sora app will initially be available for Apple’s iOS operating system, with plans for expansion to Google’s Android in the future. The updated Sora software, Sora 2, will also be accessible via the sora.com website.
OpenAI launched a new social app called Sora for creating and sharing AI-generated videos.
The app allows users to create AI avatars and voices of themselves for video integration.
Sora aims to improve AI video quality, including adherence to physics and reliable prompt following.
Source: SCMP
