Gabrielle Pesantez
On December 11th, 2025, Disney announced a landmark deal with OpenAI that would allow the emerging technology company to license iconic Disney characters for use in its Sora platform—OpenAI’s generative AI video model.1 The deal, the first of its kind in Hollywood, comes during a moment of immense legal strife between rightsholders and AI developers. This past year has seen major litigation between creatives who hold copyright in their works and AI companies who unilaterally appropriate those works for use in training their models, without first seeking authorization from rightsholders. Whether or not such a use qualifies for the “fair use” exception to copyright infringement has hinged on judges’ conflicting interpretations of the transformativeness of such training uses and doubt over whether the conduct presents a requisite level of harm to the market for the original work.
Historically, courts have limited the fair use market harm analysis to traditional, reasonable, or likely to be developed licensing markets.2 AI developers are increasingly exploring licensing agreements for content, with the Copyright Office explaining that such “markets exist or are ‘reasonable’ or ‘likely to be developed.’”3 Such a market certainly exists now for creative enterprises, and it encompasses arrangements beyond your traditional licensing scheme. Disney’s deal is structured as a joint creative venture with OpenAI’s short-form video platform.4 Though still subject to Board approval, Disney plans to purchase a one-billion-dollar stake in OpenAI. As part of the deal, “a curated selection of videos made with Sora will be available to stream on Disney+ … giving the streaming service a foothold in a type of content that younger audiences … enjoy viewing and that has proved powerful for competitors like YouTube and TikTok.”5 Sora users will be able to generate endless, 30-second video clips with Disney characters, albeit with restrictions on what activities those characters can engage in (barring inappropriate adult themes or interactions with characters owned by other media companies).6 Overall, it represents a turning point for a creative media company with immense cultural capital aiming to preserve its intellectual property arsenal where copyright protections have now come under existential threat.
The deal also encompasses only characters, with no explicit mention of human talent. Disney’s CEO Robert Iger emphasized the importance of these controls, signaling that such licensing agreements between creative enterprises and AI developers can offer a pathway toward preserving intentionality and careful involvement.7 Of course, Disney is a company with the sort of cultural capital sufficient to negotiate such a favorable agreement. Even in spite of its status and ability to secure such a deal, it is fighting other developers in court, including the AI image generator Midjourney, over its alleged infringements made on fair use grounds.8 Disney’s lawyers have also sent letters to Google, Meta, and Character.AI demanding that the companies cease using Disney’s copyrighted works to train generative AI models. Still, this deal represents a creative path forward – and reflects a worldview where creative enterprises are not simply shut out of the AI-copyright infringement conversation. Wholescale acceptance of the fair use defense in the emerging technology context, with no recognition of innovative theories like market dilution and with limited recourse for creators, would remove bargaining power from creators’ hands. Disney’s deal, backed with the adequate capital and notoriety to leverage demand for licensing of their iconic characters while staying on top of creative markets in the face of AI’s impending encroachment, is a move toward not only self-preservation, but creative domination.
[1]Brooks Barnes & Cade Metz, Disney Agrees to Bring Its Characters to OpenAI’s Sora Videos, N.Y. Times, https://www.nytimes.com/2025/12/11/business/media/disney-openai-sora-deal.html.
[2] Am. Geophysical Union v. Texaco Inc., 60 F.3d 913 (2d Cir. 1994).
[3] U.S. Copyright Office, Copyright and Artificial Intelligence, Part 3: Generative AI Training 70 (2025).
[4] Barnes, supra note 1.
[5]Id.
[6]Id.
[7]Id.
[8]Id.