Company’s new social media app allows users to create short videos with audio from text prompts and insert themselves into AI-generated scenes By Keach Hagey and Gareth Vipers
OpenAI is squaring up to TikTok, Google’s YouTube and Meta Platforms with a new social-media app for its AI video generator that allows users to create high-definition video clips with audio from text prompts.
Users can upload short clips of themselves and insert them into Sora-generated worlds, describing the idea, style and scene they want to see. They can also connect with other users, watching and commenting on their content.
The new version, Sora 2, will feature a swipe-and-scroll navigation similar to that of platforms like TikTok, Instagram Reels and YouTube Shorts, setting out OpenAI’s stall as Silicon Valley steps up its focus on AI video generation. The company plans initially to release the app through Apple’s App Store in the U.S. and Canada on an invite-only basis.
OpenAI faces stiff competition from Google, which recently connected its Veo 3 AI video generator to its popular YouTube platform, allowing users to incorporate the technology in short-form videos. Social-media and video-sharing apps are competing fiercely for user engagement.
Sora 2 will include a vertical feed and algorithm-driven recommendations that give priority to content that users might connect with, the company said Tuesday. Sora was first released in December, allowing users to create high-definition video clips from text prompts.
Technology and social-media companies are betting that new AI features will increase engagement and the popularity of their apps and services.
OpenAI’s new app joins a crowded field. TikTok’s AI Alive feature lets users turn pictures into videos with prompts, and users can upload AI-generated content. Meta last week rolled out a new feed of short-form AI-created videos in its AI app.
In an attempt to prevent doomscrolling, OpenAI said the new app won’t allow users under the age of 18 to have the infinite scroll function by default and will nudge adult users toward creating content if it perceives they have been passively viewing for too long. Content will be marked as AI generated when it is moved off platform so that its provenance is clear.
AI companies have taken an aggressive approach to how their fast-evolving tools use creative works both for training and in response to user prompts.
The Wall Street Journal’s parent company, News Corp, has a content deal with OpenAI.
The new version of Sora can create videos featuring copyright material unless copyright holders opt out of having their work appear, the Journal reported Monday.
“I think they are certainly opening themselves up to lawsuits in particular cases,” said Mark Lemley, professor at Stanford Law School, who represented the AI company Anthropic in its recent copyright case. Anthropic agreed to pay at least $1.5 billion to settle a copyright-infringement lawsuit over its use of pirated books to train large-language models.
Write to Keach Hagey at Keach.Hagey@wsj.com and Gareth Vipers at gareth.vipers@wsj.com
