The Annecy International Animation Film Festival, held annually in the French Alps, became the epicenter of a fervent debate this June as the animation world grappled with the rapid advancements of artificial intelligence. Despite a severe heatwave that pushed temperatures close to 40 degrees Celsius, animators, producers, and executives converged to discuss the technology reshaping their industry faster than anything in a generation.
A panel titled “Animation: More Human than Ever,” moderated by veteran computer-graphics educator Mark Flanagan, brought together key figures including Henry Daubrez of Google Labs, Jade Hautin from Frogbox, American technologist Benjamin Michel, and producer Leo Neumann. The central question revolved around how animation can retain its human element as creation tools become increasingly automated.
Henry Daubrez championed AI’s potential for accessibility, arguing it could empower creators in countries lacking traditional studios or software. However, he cautioned against lazy use, emphasizing that the trick is to imbue the machine with a human point of view, rather than seeking one within it. Daubrez advocated for “hybrid production,” where AI handles rendering while humans maintain control over movement and design.
Benjamin Michel focused on the economic implications, envisioning a future where small, $5 million studios produce films that once required $50 million productions, forcing larger houses to cut “padding.” Michel’s memorable line, “what’s left is you,” underscored the idea that once technology handles the technical craft, human taste and vision become paramount, repeatedly bringing the conversation back to authorship and creative control.
However, the optimism was tempered by significant anxieties. Mark Flanagan highlighted the awkward reality that while established filmmakers might use AI to realize passion projects lacking funding, younger artists fear for their first job opportunities. Producer Leo Neumann was blunt about efficiency claims, stating that for small teams, they would have been faster without AI. Jade Hautin, whose collective has tested AI tools in real productions for two years, captured the industry’s ambivalence: “Part of you wants it to work,” she said, “and part of you doesn’t.” The panelists ultimately agreed on one certainty: no one could predict AI’s trajectory in the next three years.
Step outside the tent, and the same conversation went quiet. AI is animation’s open secret. It is in almost everything now, but saying so has become a matter of nerves. Everyone wants to be the first to do something startling with it, and almost no one wants to admit they are using it at all, because of what happens when you do.
Weeks before the festival, Amazon MGM Studios and Amazon Web Services launched a fund for AI-made shows, green-lighting three projects for Prime Video, including “Punky Duck” from Mexican director Jorge R. Gutiérrez, known for “The Book of Life” and “Maya and the Three.” Gutiérrez, a vocal advocate for animators who had warned in 2024 that AI could remove the ladder for junior artists, faced brutal criticism and threats to his family. Within two days, he withdrew from the Amazon AI program, stating, “Actions speak louder than words,” and apologized to those he felt he had let down.
Gutiérrez’s ethical guidelines for AI use are clear: maintain creative control and avoid copyright infringement. He also raised a deeper concern about ownership, likening typing a prompt to hiring a stranger online, where the output isn’t truly yours. His studio learned the hard way at an Annecy test screening, where crediting AI tools led to immediate audience rejection, while studios that remained silent avoided trouble until discovered. He compared it to music, where a piece loses value upon discovering a machine, not a person, performed it.
For the people still trying to get in, the worry is simpler. The Mexican animator Quique Gasca left animation school not long ago, and the thing that keeps him up is a mechanism. AI is coming first for the in-between frames, the drudge work that has always been how a junior learns the job and how older animators pass down what they know. Removing this bottom rung, he argues, eliminates the career ladder. Gasca also fears that AI models, having “swallowed everything,” make it harder for new artists to find their unique “sound.” In response, he and his junior colleagues are focusing on tasks machines cannot replicate, such as stop-motion and the use of real materials and mistakes, though he fears this handmade route will remain a niche while cheap AI becomes the industry’s “fast food.”
Jade Hautin, while a panelist, views the situation from a different perspective. Her company, Frogbox, does not use generative AI, a stance shared by many French studios. She is also an ambassador for Creative Machines?, a French-speaking collective founded in late 2023. With over 1,100 members, the collective questions AI technology, tests tools in real productions, and hosts discussions with sociologists, lawyers, and economists.
Hautin noted the fierce polarization in AI discussions, with two hardened camps: believers who embrace AI daily and those who consider even discussing it a betrayal. Her collective, attempting to stand in the middle, faces criticism from both sides. She highlighted the rapid improvement of AI tools, citing that an AI character couldn’t blink in April 2024, but by April 2026, the results were stunning. This fear, she believes, drives AI discussions underground, as evidenced by her collective’s international think tank where participants felt safe to discuss AI for the first time.
Hautin clarified that the real issue isn’t assistive tools, which have long been part of technical pipelines, but generative AI built on scraped data, which she called environmentally “monstrous,” especially given the heatwave. She suggested that the industry’s inability to argue effectively stems from a lack of precise definition for “AI.” She differentiates between generative tools deep within specialist pipelines, which have existed for years, and browser-based tools trained on work whose authors never consented to their data being used. The fight is really about the second; models trained on work their authors never agreed to hand over. Name the thing precisely, she suggests, and the industry can finally argue honestly.
Meanwhile, in another corner of the festival, a group of Spanish and Italian creatives discussed the environmental cost of the technology. They expressed their love for their work, but questioned its worth if its future depended on an environmental disaster, a point on which they unanimously agreed.
