News
Maggie Gyllenhaal Scraps AI Face Swap of Dakota Johnson as Marilyn Monroe

The director revealed at the Venice Film Festival that her short film about Marilyn Monroe was meant to use AI to transform Dakota Johnson's face into the screen icon, but she scrapped the idea after finding the result lifeless.
Contents
Maggie Gyllenhaal tested artificial intelligence while producing a short film about Marilyn Monroe, then abandoned the technology after seeing the results. The director spoke about the experiment during a symposium at the 83rd Venice International Film Festival, where she is serving as jury president.
The idea was simple on paper: instead of makeup, prosthetics, or traditional special-effects work, Dakota Johnson's face would be digitally transformed into Marilyn Monroe's features using an AI model trained on licensed footage of the actress. Companies specializing in deepfakes and synthetic media increasingly offer this kind of technology, marketing it as a cheaper, faster alternative to classic filmmaking methods.
Why the project was scrapped
Gyllenhaal described the result bluntly: the technology wasn't up to the task. When AI-generated shots were placed side by side with Johnson's authentic performance, the gap overwhelmingly favored the human actor.
What came through in Dakota's performance as Marilyn Monroe was so much more moving, human, alive. It fundamentally did not work - Maggie Gyllenhaal, director
According to the director, the AI-generated version killed the very thing that made the film worth making. She stressed that all the reasons the crew had taken on the project in the first place disappeared in the digital version.
Who was behind the idea
The initiative didn't come from Gyllenhaal herself. The project's producers asked her to test the technology, citing licensed software trained exclusively on official, authorized footage of Marilyn Monroe, meant to set the approach apart from illegal deepfakes built on stolen data.
Still, the director made no secret of her skepticism toward the talk of "democratizing filmmaking" that accompanies the marketing of such tools. She pointed out that it's the tech companies controlling access to AI models, not filmmakers or audiences, who stand to gain the most from lower production costs.
Money and the environment
Gyllenhaal also raised a financial concern that goes beyond the Johnson experiment. She warned that if films worth tens of millions of dollars could be shot for a fraction of that cost thanks to AI, traditional film-financing models could disappear, taking with them jobs for crews, makeup artists, and supporting actors.
The second issue she raised was the environmental footprint of generative AI. The director referenced the debate over data centers being built in South Memphis, suggesting the film industry should speak openly about the energy and water consumption tied to training and running such models, instead of treating them as a neutral editing tool.
Festival context
The comments came during a symposium on Venice's Lido, where Gyllenhaal chairs the international jury for the festival's 83rd edition. She is the third woman to head the jury in five years, after Isabelle Huppert in 2024 and Julianne Moore in 2022. Earlier in the same edition, the director also criticized the low number of women among filmmakers in the main competition.
The "Flesh Impact" story made headlines in Poland as an example of a high-profile but unsuccessful AI experiment in film production, feeding into a broader debate about the limits of using artificial intelligence to recreate the likeness of deceased actors. Earlier attempts to digitally resurrect stars, including Marilyn Monroe herself in chatbot form, drew legal and ethical criticism over image rights.
For the film industry, a statement from a respected director who personally tested and rejected the technology, rather than criticizing it from the outside, carries different weight than typical objections. Gyllenhaal did not reject the idea of experimenting with new tools altogether, but she drew a clear line: the result has to hold up artistically, and for now, digital face-swapping hasn't passed that test.

