The conversation about the role of artificial intelligence in documentary filmmaking flared up again on November 15, when filmmakers and specialists gathered to examine both its promises and its ethical dangers. The discussion was moderated by Elizabeth Klinck, who opened the session by acknowledging that AI contains “the good, the bad, and the ugly,” but encouraged participants to focus on the positive potential it can bring to non-fiction cinema. Among the speakers were prominent figures such as David France and Marc Isaacs, both present at IDFA 2025 with new work, as well as Eugen Bräunig and the festival’s Guest of Honor, Susana de Sousa Dias. From the outset, France set the tone by recalling the challenge of protecting the subjects of his documentary Welcome to Chechnya (2020), a film depicting an underground network devoted to rescuing LGBTQ+ people persecuted in the Russian republic. To prevent reprisals, the team used an AI-based technique that allowed volunteers from New York’s queer community to “lend their faces” to those in danger—a complex process due to the extreme risks the sources faced.

Bräunig, a collaborator on the project, explained how the tool worked: through machine learning, it overlaid the donors’ faces onto those of the protagonists, imitating their expressions as if it were a glove adapting to the original anatomy. Although the technique resembles deepfakes, the filmmakers stressed that the process followed a strict ethical protocol, including multi-stage informed consent and clear explanations to audiences about the use of digital interventions. France noted that even though the film merely stated that people had been “digitally disguised,” the faint halo surrounding the substituted faces worked as a visual signal: there was deliberate manipulation meant to protect lives, not deceive viewers.

This concern for audience trust was taken up again by Bräunig when explaining why the Archival Producers Alliance published its Best Practices for the Use of Generative AI in Documentaries in 2024. He argued that the proliferation of synthetic images—now surpassing real ones in volume—forces creators to establish their own standards in what he described as a “Wild West” landscape. In his remarks, he emphasized the need for documentarians to adopt an ethical commitment to their audiences and to their craft, imposing standards that preserve the integrity of storytelling and maintaining transparency when AI intervenes in image construction. Rather than a rigid rulebook, the guidelines offer questions, warnings, and strategies for clearly communicating the use of generative technologies.

Susana de Sousa Dias expanded on this point, warning that AI tools have entered the documentary field with the promise of “filling the gaps of visual memory” and even reconstructing “images, voices, and bodies that were never recorded.” A specialist in archival work and in exploring what images obscure or erase, she cautioned that such recreations risk compromising the documentary status of the material and its value as testimony. In her view, the technology may lead some viewers to accept false images as true or, conversely, to distrust all visual evidence altogether. Both extremes, she said, shake the “truth regime” that documentary relies on and compel us to ask whether each technical advance will reinforce systems of extraction and exploitation, or whether it might serve as an opportunity to imagine new frameworks for thought.

To close, Marc Isaacs turned toward the creative possibilities of AI in hybrid cinema, sharing his experience with Synthetic Sincerity, a project that explores whether an AI-generated character can learn authenticity. In collaboration with Romanian actress Ilinca Manolache, the filmmaker created an avatar that allowed him to investigate the boundary between human performance and algorithmic performativity—continuing his ongoing reflection on documentary truth and representation, present in earlier works such as The Filmmaker’s House. Isaacs noted that these experiments not only question the relationship between machines and people but also transform the way we look at the human face, opening a new chapter in the conversation about the limits—and the future—of documentary filmmaking in the age of artificial intelligence.