Lionsgate and Runway: Why AI Movies Are Failing in Hollywood
A year ago, film studio Lionsgate announced a partnership with Runway, a company specializing in artificial intelligence for video. The goal was to train models using Lionsgate’s extensive film library to create new scenes or even entire movies generated by artificial intelligence. Lionsgate is known for distributing major hits such as the Hunger Games, John Wick, Twilight, and Saw franchises. Yet the entire project has reached a dead end.
During the first year of the partnership, they encountered unexpected complications. According to sources, Lionsgate’s film library is too small to create a high-quality artificial intelligence model. One insider said that even Disney’s catalog would not be enough for proper training. Runway, despite being among the leaders in artificial intelligence for video, needs enormous amounts of data to produce decent content. As a result, the output lacks continuity, character consistency, and sophisticated scene direction.
Copyright and Actor-Related Obstacles
Another major obstacle involves copyright issues. Every film has numerous rights holders, and when artificial intelligence creates a video in which an actor says something they never actually said, the situation becomes complicated. A lawyer quoted in the report explains that in the film and television industry, every production has different stakeholders with rights. This area remains legally unclear, with no clear rules for compensation or consent from actors whose likenesses appear in generated content.
The partnership has therefore run into limitations that go beyond mere technical capabilities. Lionsgate wanted to use artificial intelligence to quickly create alternative versions of films, such as animated PG-13 variants that could be produced within hours. But the reality is different—productivity over the past twelve months has been minimal.
Technical Shortcomings and Competition
Runway planned to create a custom artificial intelligence model that would be used exclusively by Lionsgate. This approach proved flawed because editors need access to multiple models to achieve professional-quality results. For example, Adobe uses several different models for its Firefly, including the recently added Ray3 model from Luma AI, as well as models from Google and OpenAI. Even Google’s advanced Veo 3 model, trained on YouTube’s entire library, has its weaknesses.
There is skepticism within the industry. Many people see the hype surrounding artificial intelligence as overblown because the technology still falls short of what investors promise. The question remains whether audiences will accept movies generated by artificial intelligence at all, even if these obstacles can be overcome.
Lionsgate emphasized that its agreement with Runway is not exclusive, suggesting that the studio is exploring options with multiple artificial intelligence providers. Nevertheless, interest remains in using this technology to adapt content and reduce visual effects costs. However, a year full of problems shows that the path to fully AI-generated movies is longer than expected.
The reality of artificial intelligence in Hollywood has revealed promising ideas colliding with the harsh realities of data limitations, legal uncertainty, and technical shortcomings. While Lionsgate continues experimenting, major breakthroughs still seem far away.
