Lionsgate is learning firsthand how challenging it is to turn ambitious AI promises into reality. Earlier this year, Michael Burns, the studio’s vice-chairman, claimed that through a partnership with generative AI company Runway AI, Lionsgate could transform one of its major franchises—like John Wick or The Hunger Games—into a fully animated anime film in just hours using AI. That vision has not materialized, according to a report from The Wrap, which reveals that the collaboration has run into significant technical and legal hurdles. The deal, announced last year as a groundbreaking partnership between a major studio and an AI firm, was supposed to give Runway AI access to Lionsgate’s entire film library to train a custom, exclusive AI model. The goal was for Lionsgate to use that model to generate AI-driven content, including new scenes, visual effects, or even entire films. However, the plan has stalled. The primary issue? Lionsgate’s library, while extensive, isn’t large enough to train a robust, functional generative model capable of producing high-quality, coherent video output. This limitation underscores a fundamental truth in AI development: effective generative models require vast datasets. Even advanced systems like Google’s Veo and OpenAI’s Sora, which are trained on enormous amounts of video data, still struggle with glitches, unnatural movements, and uncanny valley effects. A model trained on a much smaller dataset—like Lionsgate’s library—would face even greater constraints, limiting its ability to produce realistic or useful results. Beyond technical challenges, the project is bogged down by unresolved legal questions. Burns had suggested that even if the studio could generate an AI version of a film, it would still need to compensate actors, writers, directors, and other contributors for its use. But who exactly qualifies for payment? The report raises concerns about whether creators, including cinematographers and production staff, must be paid when their work is used to train AI models or reinterpreted in new forms. The fact that Lionsgate owns the intellectual property doesn’t automatically resolve these issues, especially as copyright laws around AI training remain unclear and evolving. Despite the setbacks, Lionsgate maintains that the partnership is still progressing. Peter Wilkes, the studio’s Chief Communications Officer, told Gizmodo that the company remains pleased with its collaboration with Runway and other AI initiatives. He emphasized that AI is being used successfully in current projects to improve production quality, streamline workflows, and reduce costs—especially in licensing and archiving the studio’s vast library. There are signs that Lionsgate is still using Runway’s tools, but perhaps not through the original plan. In a previous report, the studio was developing an AI-generated trailer for an upcoming film that hadn’t yet been shot—using fabricated scenes to pitch the project to buyers. While this could offer marketing advantages, it also raises questions about authenticity and audience trust. Ultimately, Lionsgate’s experience highlights the gap between AI hype and practical application. While the technology holds promise, building a reliable, legally sound, and creatively viable AI system for film production remains a far more complex challenge than early predictions suggested.