The Book of Veles, photographer Jonas Bendiksen’s latest project, is a fresh and unnerving mediation on authenticity, veracity and truth – questions that have dogged photojournalism with every new advance in imagemaking. 

Ostensibly a photobook on the small Macedonian town of Veles, which made international headlines in 2016 as the unlikely factory of pro-Trump fake news, The Book of Veles created a furore after Bendiksen revealed that the project’s images were synthetically generated using 3D software and the book’s text was written entirely by artificial intelligence. 

Fascinated with the story of Veles as well as with developments in synthetic imagery, Bendiksen set out to see just how “real” he could make his images. Although he left breadcrumbs throughout the text, to his surprise the book was published in April 2021 to “nice, positive echo-chamber feedback,” said Bendiksen. No one questioned the authenticity of the images or text. He stepped up the game by then submitting his manipulated photographs to the world’s most prestigious photojournalism festival Visa pour l’image which screened his images in September.

“I thought, what could be a higher threshold for fooling someone with junk, synthetic images than this crowd?,” said Bendiksen. “I gave it 24 hours for someone to come with some questions about the work. It didn’t happen.” The photographer’s final attempt to out himself involved buying a squadron of Facebook and Twitter bots to attack him online. The bots posted dozens of messages claiming that his work was fraudulent, only to have Bendiksen’s colleagues and supporters rush to his defense.

Bendiksen finally came clean in an interview on September 17 with Magnum Photos. We spoke with Bendiksen about what he’s come away with from the experience and how the project has continued to take unexpected turns even after his revelation last month.

This conversation has been edited for length and clarity.

What were the main goals that you set out with?

I wanted to create some discussion and awareness around this technology but I didn’t go into this project wanting it to be just a technical demonstration of something. I get into my projects because it’s a good story to tell. If it hadn’t been for all these exciting layers of Veles and mythology I wouldn’t have done this. All of these parts of the puzzle fell miraculously into place.

Yes, tell us more about how the puzzle fell into place, there are so many bizarre layers to this project.

One was the discovery of the historical layers to the story. The town Veles is named after this pre-Christian Slavic god who was a kind of sneaky guy, a shapeshifter who turns into a bear, a god of chaos and deception and magic. I thought he would probably be super happy about all this deception going on in Veles.

The introduction written by AI is still a bit clumsy. Some of your images, especially the one of a bear stomping through town, feel like they should have set off some alarm bells. Yet it was all “real” enough that no one questioned it.

That was also something that frightened me. The technology in the field has developed a lot even from when I started using it to when the project was done. It’s clear to me that within a few years 95% or more of people will have a hard time decoding whether an article was written by a New York Times journalist or a bot. I wanted this to be a look into what I think is the near future of our information landscape.

How are our current concerns about automation or AI different from all the previous technology scares?

In photography at every step when there is a new technology people say it’s the collapse of truth. Whether that was when digital cameras came or when Photoshop showed up. People always cried wolf like that. Maybe I’m the same and this is a bunch of wolf crying and this will sort itself out nicely.

But I think the difference is: automation and synthetics. We will always have good journalism but I think it will be mixed in with so much synthetic junk or half synthetic junk that it will just be very, very chaotic and difficult to navigate. The difference is that the automation of it gives it such a bigger potential for spread and makes it so difficult to contain.

You’ve repeatedly compared your work to a penetration test that hackers run searching for vulnerabilities in their code. These tests then allow companies to better fortify their software and find solutions for loopholes in the code. What did your stress test reveal and what solutions did it reveal to you?

I’m not trying to pretend I have all the answers but I think there are many levels to it. I think the content verification business will be an industry unto itself. This is also a call out to social media platforms who I believe have failed us in many ways on this front thus far to step up the game. This is a call out to our education system. As a father of four, I’m wondering how my children are going to handle this. To be a functional citizen in the next 50 years, navigating the information space should be at least as important a subject as mathematics in schools.

This story continues to go in unexpected directions. Tell us about the latest.

A junk information site pretending to be a newspaper in Texas called Texas News Today came out from a very similar story that was stolen from Wired about my project. I think it’s automated: they suck stuff into their system, rewrite it and then it gets blasted out again on all these websites which look very similar to what the fake news websites out of Veles were doing. It’s the same business model. I looked into it and it turns out these people at Texas News Today are a junk news site based in Pune, India running a bunch of sites from this location in different languages.

Now, these fake news sites are stealing stories about my fake news project. And there are people quoting the fake writer who supposedly wrote the piece as some credible source. It’s full circle. There you can see the whole mechanism of the chaos.