OpenAI announced on Tuesday the closure of Sora, its AI-generated short video social network, just six months after its launch marked by great enthusiasm.
“We say goodbye to Sora,” the company wrote on X to announce this decision which confirms its refocusing on professional tools before a possible IPO later this year.
Disney, which in December granted a license to use the image of its animated characters in Sora, will withdraw from this partnership, according to a source close to the matter cited Monday by the Hollywood Reporter. In return, the studio committed to investing a billion $ in OpenAI.
This closure of Sora marks the end of one of the most publicized artificial intelligence products of last year, at a time when the parent company of ChatGPT is narrowing its focus on coding and productivity tools, a niche in which its rival Anthropic is eating up more and more ground.
OpenAI added that it would soon communicate on the timetable for stopping the application, which consumes computing power, as well as on the terms of conservation of its users’ creations.
This closure comes at a delicate time for OpenAI, whose economic model is increasingly called into question: costs are exploding much faster than revenues, despite around a billion daily users worldwide.
According to the Wall Street Journal, Sam Altman announced this refocusing to employees on Tuesday.
The announcement also comes in the wake of the message from Fidji Simo, head of applications at OpenAI, who asked his teams in early March not to be distracted by “side quests” – a term borrowed from the world of video games – to focus on AI agents.
These tools, capable of autonomously chaining tasks on users’ computers to write code, analyze data and make decisions in different applications, are now the center of attention of the Silicon Valley giants.
In mid-February, OpenAI hired Austrian programmer Peter Steinberger, the originator of OpenClaw, software for developing AI agents which has enjoyed dazzling success among computer scientists around the world.
For its part, Anthropic on Monday deployed an agentic functionality of its Claude model, now capable of controlling the user’s computer to accomplish tasks autonomously.
