Can AI give Arab filmmakers the freedom to tell our stories?

I thought I had a good eye for detecting AI-generated images, until I watched the trailer for Dreams of Violets, the Iranian movie that premiered at the Toronto International Film Festival this year. Had I not known in advance that it was made entirely with artificial intelligence, I would have assumed it was an over-polished film with real actors.

Like most filmmakers struggling for years to raise funds to make their own films, I was enthralled by the technology, but more horrified by how advanced it had become.

The fact that an artificial film was selected by a leading film festival feeds into the industry’s fear that technology will obliterate human creativity. No more actors, cinematographers, set designers, or sound recordists - everything will be made from a desktop. Thousands will lose their jobs, and thousands more will churn out more AI-generated films that the human population can actually watch.

But the idea of making a film in just a few months on a budget of only $2,000, rather than spending years trying to raise $750,000- $800,000 to make one, is undoubtedly attractive.

Most independent filmmakers spend years writing and rewriting, applying for funds, submitting endless documents explaining why their film deserves to be made. They endure one rejection after another before they win a modest grant that barely covers the cost of wardrobe - basically doing everything except actually making the film.

By the time the film finally gets made, filmmakers have outgrown their own projects, but they cannot turn back, so they plough through and pray that the film comes out well enough for them to do it all over again with another project.

As I write this, I am months away from shooting my first feature. The idea that I could avoid all of that and make my film exactly how I imagined it, entirely on my own terms, was too seductive to ignore. So I decided to run an experiment on a short scene from my script.

The scene involves a grieving couple. They’re standing outside their home in the Palestinian town of Jericho. The woman stands outside the front gate, her back turned to her husband, who stands way back at the front door. She stares at the horizon, watching the sunset apprehensively, as he looks on despondently. She hears the call for prayer, mutters a few words for herself, and then goes inside. Simple.

First, I needed to create the characters on Google’s Nano Banana. Starting with the woman, I wrote the following prompt: A Palestinian woman in her forties. She wears a hijab and a simple dress. She is beautiful but sad. It’s the kind of sadness that never leaves your face even when you smile.

In less than 10 seconds, Google produced this image:

Notice that I said nothing about the background, but it figured it out from the word “Palestinian”.

The moment I saw her face, I wanted to stop the experiment. This isn’t the face of anyone that I know, yet it had a bit of every Palestinian woman that I know. I did not provide Google with a photo reference, but how many photos had it already ingested from around the world to create something so real?

I had completely forgotten that a few weeks before writing this article, I was so concerned about the question of “creation through AI” that I contacted an Islamic institute to see if there was an advice/fatwa on the matter. No one responded, so I have to assume this experiment is somehow in the halal territory.

But the face was so real I was concerned that a woman may recognise herself in the image, but I decided to prompt Google again, using a photo of myself as a reference, so if anyone criticises me, I can at least defend the experiment by saying that I didn’t steal anyone’s face.

At first glance, the woman doesn’t exactly look like me, but I can see elements of myself in there. To keep it safe, I asked Google to create an image of the husband - also using my photo as a reference. I will leave it to the readers to judge the result…

Next, I searched for an AI platform that can take these images and create the video. Higgsfield is one such platform, and I found it intimidating. With its endless tools and complicated features, browsing through the site was a revelation; that there is a generation of artists and filmmakers is already fluent in this technology - and the rest of us had better start paying attention.

To make the experiment work, I had to buy a subscription that lets me run it multiple times to get the right result. Using images from actual buildings in Jericho, I prompted the site to create that beautiful sunset scene from my script. But instead of using the above images, Higgsfield produced a photo of a young man and woman standing outside a cave. I tried again with a firm prompt to only use the images that I provided, and the result was somewhat decent - except for the fact that it decided to give the woman a few lines to say at the end of the scene. It was very short, a mixture of Arabic, maybe some Farsi, Pashto, Urdu. I don’t know.

I decided to stop. I now know that with an advanced subscription and several days of continuous trials and prompting, I will get a better result. I went over my script, scanned through countless AI-generated videos on Higgsfield and elsewhere, and I can already see what my entire film would look like - from start to finish. Polished, technically flawless, with a diminished soul.

One of the greatest joys of film directing is working with actors - discussing the characters, trying different approaches for each scene, and then doing multiple shoots and coming up with a different result each time. No two actors are the same. Different talents and personalities require different approaches, and a good director knows how to find the unique language to work with each individual.

Sometimes, accidents happen on set which trigger an idea that elevates the film in a way no one had anticipated.

Dreams of Violets has not yet been released, so I watched the trailer again and wondered to what extent the director, Ash Koosha, really tried to achieve different results with his artificially created characters. It is worth noting that Koosha is a technology entrepreneur and the co-founder of Clairgrid - an AI start-up. But I doubt that access to advanced technologies could break through the mysteries of human behaviour and creativity that make acting so unique.

To date, most reviews of this film depict it as a bland production, with some saying that the use of AI dampened the impact of depicting the violent events of the 2026 protests. Ash Koosha said he decided to make the film with AI because he felt an urgent need to tell this story and knew he could not film in Tehran.

While the urgency may be legitimate, this project is exactly the type of film that Western funders would scramble to support. With the right amount of money and good production design, Tehran can be recreated in Morocco, Jordan or Algeria.

Nevertheless, I keep coming back to the sense of liberation one feels knowing they can make their film without waiting for anyone's approval. It’s not just the approval of money, but also of the ideas themselves.

Everywhere we turn, there are invisible red tapes. Funds, donors and governments deny their existence, but we know they are there. Europeans will only support projects from the global South if the issues and characters are relatable to them.

In the region, it is difficult to shoot anything without an official seal of approval from local authorities, and certain topics and taboos are off limits. When I was preparing to shoot my last short film in 2023, some producers and actors declined to participate at first because they didn’t want to get in trouble with their local community.

In such cases, using artificial intelligence to tell forbidden stories becomes part of the story itself. The technical and dramatic flaws of such productions become part of the conversation about a film that otherwise might never have seen the light of day.

I don’t plan to experiment with AI on my debut feature film any further. But the next project is one where I know that using this technology may, in fact, be my only option - and yet I hope it is not.

Dima Hamdan is a Palestinian journalist and filmmaker. Her latest film, Blood Like Water, was screened in over 80 festivals worldwide, winning ten Best Short Film awards. She is also the manager and editor of the Marie Colvin Journalists’ Network, an online community dedicated to supporting Arab women journalists.

Have questions or comments? Email us at editorial-english@newarab.com

Opinions expressed in this article remain those of the author and do not necessarily represent those of The New Arab.