Q: Why did Steven Soderbergh’s John Lennon documentary spark debate about generative AI in documentary filmmaking at Cannes? A: Soderbergh used about ten percent of the film’s visuals generated with Meta’s video tools and accepted Meta financing for completion, and he framed the AI segments as surreal rather than literal. Cannes amplified the debate because some filmmakers vocally opposed AI while others were cautiously open, creating mixed signals about when and how generative AI belongs in documentary filmmaking. Q: How much of John Lennon: The Last Interview used AI and what kind of shots did Soderbergh include? A: Roughly ten percent of the documentary’s visuals were AI-generated and those segments were abstract or dreamlike—circles of light, a black rose morphing, painterly diptychs, and split-screen lovers—used to score long stretches of audio. There are no AI deepfakes of John Lennon or Yoko Ono in the film. Q: What ethical test did Soderbergh apply before using an AI shot in the film? A: Soderbergh said every AI shot had to pass three questions—was it necessary, was it the best way to achieve the image, and was it the only way to do so—and he placed AI late in finishing to solve images that were otherwise impossible. He also emphasized radical disclosure and described himself as “my own whistle blower” to make the process transparent. Q: What are the five ethical rules recommended for using generative AI in documentary filmmaking? A: The five rules are: 1) purpose over novelty—use AI only when it passes the necessity test; 2) never fabricate people or events and obtain explicit consent for likeness or voice; 3) radical transparency—label AI use, cite tools and sponsors, and keep a shot log; 4) protect human craft by hiring, crediting, and paying people rather than relying solely on prompts; and 5) track provenance and respect data, privacy, and archive rights. Together these rules are meant to keep generative AI in documentary filmmaking serving truth, protecting labor, and maintaining audience trust. Q: How should filmmakers disclose AI-generated content to audiences and festivals? A: At minimum place an on-screen statement such as “This film contains AI-generated imagery,” and list tools, vendors, and the specific sequences that used AI in the end credits. Productions should also disclose sponsorship or funding conflicts and keep an internal shot log marking source type and timecodes to share with press, festivals, and fact-checkers. Q: Is it acceptable to create deepfakes of historical figures in documentaries? A: The guidance is clear: do not create deepfakes of real people without explicit, verifiable consent from the individual or their legal representatives. If estates or families approve digital performance work, that consent should be stated on screen and in credits, and filmmakers should avoid generating images that viewers could mistake for authentic historical footage. Q: How can productions protect jobs and craft when using AI tools? A: Use AI to extend human teams rather than replace them by hiring researchers, editors, VFX supervisors and other crew first and crediting and paying them accurately. Soderbergh argued that most filmmaking jobs cannot be performed by the tech and that as technical perfection becomes easier, human imperfection and craft become more valuable. Q: What should festivals do about films that use generative AI in documentary filmmaking? A: Festivals should nudge best practice by requiring simple on-screen labels like “contains AI-generated imagery” and submission checklists for provenance to raise the bar without shutting doors. Cannes 2026 showed mixed signals—organizers discussed excluding films “primarily driven by generative AI” from competition and even floated a visible “made without artificial intelligence” label, but that policy was not codified in the published regulations.