OpenAI is negotiating with a startup over deep access to Sora, its text-to-video generative AI model. The goal is to build AI video tools designed specifically for filmmakers.
The startup has not been named. The deal terms and timeline have not been disclosed either, which is standard for talks still in progress.
What’s on the table
The talks center on a third party getting privileged access to Sora so it can build specialized products on top of it. Those products would serve people who make films for a living, not casual users who want a cat riding a skateboard through Tokyo.
The arrangement also fits a pattern. OpenAI has been pursuing selective partnerships around Sora rather than simply opening the floodgates. It has previously held discussions with major studios, including Disney, Universal, and Warner Bros., about custom versions of Sora for their own internal use.
How Sora got here
OpenAI first previewed Sora in February 2024. At the time, access was limited to a select group of researchers and artists who could test the model and provide feedback.
The public launch followed in late 2024. Sora arrived positioned as a cutting-edge tool for generating video from text prompts, part of a wave of generative AI that moved from writing essays to producing moving images.
Since the initial announcement, various Hollywood factions have raised concerns about copyright, rights to likeness, and fair compensation for creators whose work may surface in the AI’s output.
OpenAI has responded by adjusting Sora’s content controls and exploring licensing frameworks. It has kept pushing toward integrations and enterprise collaborations at the same time.
What this means for Hollywood and AI video
For the film industry, a startup-built tool layer on top of Sora could change how AI reaches working creatives. Studios negotiating custom versions for internal use is one path. A specialized startup selling tools to filmmakers more broadly is another.
For OpenAI, the strategy has a clear logic. Partnering with a startup lets it reach a specialized market without building every filmmaker-facing feature itself. The startup handles the workflow details, and OpenAI supplies the underlying model.
A few questions are worth tracking as negotiations continue. The first is the startup’s identity, which would reveal what kind of company OpenAI trusts with privileged model access. The second is what “deep access” actually means in practice, whether that involves customization, priority capacity, or something else entirely.
The third is how any resulting product handles the copyright and likeness concerns that have trailed Sora since February 2024. Reports emerging from early 2026 indicate that OpenAI is shifting its focus regarding Sora, with deployments of app features and adjustments in partnerships expected as negotiations progress.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.