Disney Could Have Sued. Instead, It Chose a $1 Billion Handshake with OpenAI

In December 2025, The Walt Disney Company struck a three-year licensing agreement granting OpenAI’s video-generation platform Sora access to more than 200 characters from Disney, Pixar, Marvel, and Star Wars. The compensation was not a traditional licensing fee. Disney took a $1 billion equity stake in OpenAI, plus stock warrants tied to future valuation growth. The world’s most powerful content owner chose to become a shareholder rather than a plaintiff.

The deal’s rights architecture is deliberately precise. Visual elements of characters are licensed for video generation on Sora and still-image generation on ChatGPT, while talent likenesses and voices are explicitly excluded. This separation of character IP from personality rights sets a structural precedent for entertainment-industry AI licensing. Disney also becomes a major OpenAI customer, deploying APIs across its products and rolling out ChatGPT for internal use, with selected fan-generated AI content slated for integration into Disney+.

The agreement’s significance is inseparable from its context. More than 70 lawsuits worldwide now challenge the unlicensed use of copyrighted works in generative AI training data. Anthropic faced a $1.5 billion class-action settlement (Bartz v. Anthropic), with Judge William Alsup of the Northern District of California ruling that the company’s downloading of over 7 million pirated books was “inherently, irredeemably infringing.” Disney itself has sued Midjourney, calling it a “bottomless pit of plagiarism.”

Against this backdrop, Disney’s approach to OpenAI represents a calculated pivot from courtroom to boardroom. The Kluwer Copyright Blog characterizes the deal as “private ordering”—bilateral contracts stepping in where statutory copyright law has yet to offer clear rules on AI training data. With fair-use doctrine still in flux across jurisdictions, contractual licensing reduces uncertainty for both parties. Disney secures a concrete economic return through equity, while OpenAI eliminates litigation risk and gains legitimate access to some of the most recognizable IP on Earth.

Yet this private-ordering model has structural limitations. It privileges large rights holders with bargaining power while offering independent creators little leverage. The Writers Guild of America East has objected that companies like OpenAI have “stolen vast libraries of works owned by the studios and created by WGA members and Hollywood labor to train their artificial intelligence systems.” A $1 billion equity position is available only to Disney-scale entities, and the deal cannot serve as a universal template. As Devlin Hartline of the Hudson Institute notes, “Arts and entertainment are not luxuries; they are part of the cultural backbone of society. When incentives to create are eroded, everyone loses—including AI companies.”

The agreement also carries implications for content industries outside the United States. Japan’s Copyright Act Article 30-4 has permitted relatively broad use of copyrighted works for machine learning, but Disney’s strategy operates on a different plane: building a commercial model that ties content value directly to AI platform growth. Major Japanese IP holders—Nintendo, Kodansha, Shueisha—could explore similar equity-for-licensing structures for their character portfolios. Nintendo, already actively enforcing its IP in the Palworld litigation, exemplifies the strategic choice between litigation and licensing that every major content owner now faces.

What Disney has demonstrated is that the resolution of AI copyright disputes need not be confined to the courtroom. Equity-based profit sharing, granular rights segmentation (characters separated from likenesses separated from voices), and embedded governance conditions (age-appropriateness, harmful-content prevention, rights-holder protections) form a composite licensing model that functions—at least for major rights holders—as a viable alternative to litigation. The critical question ahead is where this model’s reach ends: whether it can extend to independent creators, and how it interacts with copyright regimes in Japan, the EU, and beyond.