A U.S. federal court has rejected Chinese artificial intelligence startup MiniMax’s bid to dismiss a copyright infringement lawsuit filed by major Hollywood studios, including Disney. The decision, handed down this week, allows the case to proceed to discovery and potential trial, marking a significant moment in the escalating legal battles over whether AI companies can legally train their models on copyrighted creative content without permission or compensation.
MiniMax, a Beijing-based generative AI firm, was sued by major studios last year for allegedly using copyrighted films, television shows, and characters to train Hailuo, its video generation AI system. According to the studios’ complaint, MiniMax not only scraped protected material for model training but also used iconic Hollywood characters and imagery in marketing materials, explicitly positioning Hailuo as “a Hollywood studio in your pocket.” The lawsuit represents one of the most direct challenges to date against Chinese AI companies accused of large-scale copyright violations—a charge that has increasingly become a flashpoint between Silicon Valley, Hollywood, and China’s rapidly advancing AI sector.
The court’s decision to deny MiniMax’s motion to dismiss is legally significant because it signals skepticism toward the company’s core defense arguments. MiniMax had sought to end the case early by arguing that its use of the copyrighted material fell under fair use protections, a legal doctrine that permits limited use of copyrighted works for purposes like criticism, commentary, or transformation. The rejection of this argument at the motion stage suggests the judge found sufficient merit in the studios’ allegations to allow the case to proceed. This sets a potential precedent that could influence similar lawsuits pending against other AI companies, both Chinese and American, facing similar accusations.
The implications for the global AI industry are substantial. Unlike previous copyright disputes in the digital age—such as those involving search engines or social media platforms—this case directly challenges the business model of generative AI development itself. Training large language models and video generation systems requires massive datasets, often sourced from the internet without explicit licensing agreements. Companies like OpenAI, Google, and Meta face parallel copyright suits in the United States and Europe, but the MiniMax case is notable for involving a Chinese firm without U.S. operations, raising questions about jurisdictional enforcement and international intellectual property standards.
For India and South Asia’s emerging AI ecosystem, the outcome carries strategic weight. Indian AI startups and research institutions are increasingly developing foundation models and generative AI systems, often using data sourced from the internet. The legal precedent established in this case—whether copyright holders can sue for model training without explicit licensing—will influence how Indian companies approach data acquisition and model development. Several Indian AI firms have already begun negotiating licensing agreements with content creators and copyright holders, partly in anticipation of tighter legal standards. The case also highlights the competitive advantage that companies willing to engage in more legally cautious practices may gain over time.
The studios’ allegations specifically named the use of copyrighted characters and intellectual property in marketing as a particularly egregious violation, going beyond mere training data incorporation. This distinction matters legally and strategically. Using protected characters to promote a commercial product crosses from the ambiguous territory of AI training into clearer trademark and publicity rights infringement. MiniMax’s marketing approach—positioning Hailuo as capable of recreating Hollywood-quality content using famous characters—essentially advertised the infringing capability itself, strengthening the studios’ case considerably. This dimension may prove crucial in influencing how other AI companies market their generative capabilities going forward.
The next phase will involve discovery, during which both sides will exchange documents and evidence. This process typically reveals critical details about how MiniMax sourced training data, what contractual arrangements existed with data providers, and internal communications about the company’s awareness of copyright concerns. Industry observers are closely watching whether MiniMax will settle the case or fight it to trial, as settlement discussions often shape broader industry norms around licensing and compensation. The outcome could force a reckoning: either the AI industry must develop robust licensing frameworks for training data, or copyright holders will need to accept that model training is effectively fair use. A verdict favoring the studios would likely accelerate the shift toward licensed datasets and potentially increase the operational costs of developing competitive AI systems globally.
For now, the rejection of MiniMax’s motion to dismiss represents a turning point. It signals that courts may not accept broad fair use defenses for generative AI training without careful scrutiny of the specific facts and uses involved. As this case progresses through discovery and toward potential trial, its trajectory will likely influence how Chinese AI companies operate, how Indian startups approach data governance, and ultimately what legal and commercial frameworks emerge to govern AI development in an age of strict intellectual property enforcement. The stakes extend far beyond one company’s legal troubles—they touch the fundamental question of whether generative AI, as currently developed, can coexist with traditional copyright law.
