Can AI Be Used in Documentary Production: The Ethical Implications?

Artificial intelligence is the most significant development in documentary production in decades. It is also one of the most misunderstood. The conversation tends to veer quickly toward either uncritical enthusiasm or reflexive concern, and neither of those serves the people trying to make good decisions about their films.

Here is an honest account of what AI can and cannot do in documentary production, and why I have integrated it into my workflow.

An AI scene from my recent feature length Documentary about the island of Madeira

How AI is Changing Documentary Production

The area where AI has genuinely changed what is possible is the recreation of scenes, moments, and environments that can be costly and challenging to capture on camera. A historical event from a century ago. An underwater environment that would cost a fortune to shoot. An abstract concept made visual. An archival photograph brought to life to create a historically accurate scene.

Before AI, a documentary that wanted to show the founding of a city, the crossing of an ocean, or the interior of a building that no longer exists had two options: expensive live-action recreation with period costumes and sets, or static photographs with slow zooms. It offered a choice between a prohibitive cost or a mediocre depiction that fell short of the scene's full potential.

AI-generated imagery, used carefully and transparently, opens a third option. It is honest about what it is, never presented as authentic archival material. It brings a scene fully to life rather than settling for a flat image or a slow zoom. And it costs a fraction of what live recreation would require.

When AI Should Not Be Used in Documentary Filmmaking

AI has no place in the representation of living people without their explicit consent. It should not be used in the fabrication of events that did not happen, presented as if they did. It would be wrong to use it to replace actual interviews that convey the authentic human moments that are the heart of documentary filmmaking.

A documentary derives its power from truth. The moment an audience suspects that what they are watching has been manufactured rather than captured, that power collapses. AI used ethically enhances the storytelling around the truth. It does not replace the truth itself.

The 2024 Netflix true-crime documentary What Jennifer Did, about a 2010 Canadian murder-for-hire case, ran into exactly this problem. Around the 28-minute mark, the film shows two photos presented as authentic snapshots of subject Jennifer Pan from her youth, illustrating a friend's description of her as bubbly and genuine. Viewers quickly noticed the images had the telltale signs of AI generation, mangled hands and fingers, a misshapen gumline, distorted background objects. The photos were never disclosed as AI-altered anywhere in the film or its marketing.

The backlash was immediate and widespread. Viewers said the discovery made them distrust not just this documentary but the entire true-crime genre going forward. One of the film's executive producers later told a reporter that Pan's face in the image hadn't been altered, only the background, done to protect the identity of the person who'd supplied the original photo. Whether or not that explanation is the full story, it came only after viewers caught the manipulation themselves, not as something the filmmakers disclosed upfront.

That's the same failure that sank the 2021 documentary Roadrunner, about chef and author Anthony Bourdain. Its director used AI to generate three lines of narration in Bourdain's own voice, words Bourdain had written but never actually spoken aloud, built from roughly a dozen hours of his recorded speech fed into a voice model. The AI narration wasn't mentioned anywhere in the film. It only surfaced after release, when a journalist noticed one line sounded slightly off and questioned the director about it. The reaction was swift: critics who'd already published favorable reviews felt misled, and Bourdain's own widow publicly disputed the director's claim that the estate had approved it.

Both cases share the same root problem. A documentary asks its audience to trust that what's labeled as historical record actually is historical record. The moment that trust turns out to be misplaced, even over something as small as two photographs or sixty seconds of audio, it doesn't stay contained to that one moment. It spreads backward over everything else in the film, and often outward to every other documentary carrying the same distributor's or genre's name.

How AI Improves Documentary Pre-Production

In my workflow, AI is most valuable during the pre-edit phase, before any shooting begins. It allows me to visualize scenes that would otherwise require waiting for a shoot, to test ideas about how historical or abstract material might work in the cut, and in some cases to include content in the finished film that would have been prohibitively expensive to produce any other way.

The result for clients is a richer film for a leaner budget. Not a compromised film. A more ambitious one.

Why Transparency Matters When Using AI in Documentaries

My practice is to be fully transparent with clients about what in a film was AI-generated and what was captured. In films where AI imagery appears, I believe the audience should understand what they are seeing. This is not a legal requirement yet in most contexts. It is an ethical one, and one I take seriously.

Next Steps for Your Documentary Project

If this article raises questions about your own situation, working through my guide, Finding Your Story, may be a useful next step. Available free at jdfreedman.net, it is designed to help organizations think through the same story, audience, and production questions I discuss with clients during the early stages of a production.

If, after reviewing the guide and completing its interactive worksheet, you would like to discuss what you discovered, you are welcome to share your results and schedule a complimentary thirty-minute consultation.

No obligation. Just an opportunity to run your ideas past someone who's been down this path before.