Meta AI powers 10% of Soderbergh's Lennon doc at Cannes; Aronofsky teams with DeepMind
Two of cinema's most innovative directors are using AI not to generate films from scratch, but to meticulously reconstruct lost moments—forever changing documentary ethics and the business of filmmaking.
At the 2026 Cannes Film Festival, Steven Soderbergh's documentary 'John Lennon: The Last Interview' relies on Meta AI for roughly 10% of its imagery—specifically to reconstruct the appearance of Lennon and Yoko Ono during their final radio conversation. This is not a film generated by AI; it is a film that selectively uses AI as a restoration tool. The move quietly highlights a fundamental shift: AI in cinema is moving from wholesale generation to precise, guided reconstruction. Meanwhile, Darren Aronofsky's studio Primordial Soup has partnered with Google DeepMind to explore AI-assisted filmmaking, signaling that even auteur directors are now embedding these tools into their creative workflows.
The competitive landscape reflects a race to serve this emerging demand. OpenAI's Sora offers broad text-to-video generation for previsualization and short clips, but faces scrutiny over control. Runway's Gen-3 Alpha provides real-time editing tools tailored for professional production, giving filmmakers granular control. Stability AI's open-source Stable Video Diffusion prioritizes accessibility and customization. Meta's involvement in Soderbergh's project is more targeted—image reconstruction rather than full video generation—but the prestige of a Cannes documentary positions its tools for premium licensing. Industry data projects the generative AI in film market will reach $1.5 billion by 2028, growing at a 35% CAGR, making these early partnerships crucial for both tech companies and studios.
The deeper implication is a redefinition of authenticity. The use of AI to recreate deceased subjects—as seen in the 2021 'Roadrunner' documentary with Anthony Bourdain—raises unresolved ethical questions about consent and historical accuracy. Soderbergh's 10% figure may actually undercount the AI contribution if background elements or rotoscoping are included, creating a subtle but critical trust issue. Without industry-wide disclosure standards, audiences may be unknowingly exposed to reconstructed reality. Aronofsky's partnership with DeepMind suggests a future where AI becomes a standard tool in the director's kit, similar to how digital color grading or CGI transformed cinema earlier. But unlike those tools, AI reconstruction touches the very nature of indexical truth in documentary filmmaking.
The bottom line: Soderbergh and Aronofsky are not using AI to replace human creativity—they're using it to augment specific, often delicate tasks. This narrower application may be the key to widespread adoption, as it sidesteps the existential panic around full AI filmmaking. However, it also introduces new legal and ethical frontiers, such as the unresolved questions around the Lennon estate's potential objections. For filmmakers and studios, the lesson is clear: AI is becoming a routine part of production, but without clear norms, each project risks becoming a flashpoint for debate over what constitutes a real documentary.
- Soderbergh's 10% AI imagery for a historical reconstruction highlights a trend: AI as a precise restoration tool rather than a generator of entire films.
- The market for generative AI in film is projected to reach $1.5B by 2028 (35% CAGR), with companies like Meta, DeepMind, OpenAI, Runway, and Stability AI competing for premium and professional niches.
- Ethical risks around consent and truthful representation remain unresolved, echoing the 'Roadrunner' controversy; future projects will likely require new disclosure standards to maintain audience trust.
Why It Matters
AI is entering cinema not as a replacement but as a nuanced tool—forcing a renegotiation of authenticity, consent, and creative control.