“Through the Lens of Deception: A Photographer’s AI-Fueled Controversy Sparks Fan Backlash”

Following Russia’s incursion into Ukraine, which disrupted his travel itinerary, Belgian photographer Carl De Keyzer opted for a virtual journey to Russia instead.

From his residence, the esteemed documentary photographer commenced working on a series of visuals depicting Russia, utilizing the capabilities of generative artificial intelligence (AI). He was taken aback by the aftermath.

During the late 1980s, De Keyzer visited Russia a dozen times in just one year. The USSR was on the brink of collapse, and De Keyzer documented the customs and leisure activities that would soon vanish. He returned in the 2000s to capture scenes within Siberia’s prison facilities.

In November, three decades after his initial voyage to Russia, De Keyzer released a collection of AI-generated visuals in a publication titled Putin’s Dream. This time, there were no human forms, no captured moments, but rather a concept brought to existence via computers.

Hours after he shared online regarding Putin’s Dream, De Keyzer encountered backlash for having crafted fictitious images and potentially adding to the spread of misinformation.

By August 2023, an estimated 15 billion images utilizing text-to-image algorithms — a category of artificial intelligence where textual prompts instruct software to generate new visuals — had already been produced. 

With the rise of generative AI imagery, ethical concerns are escalating. This has turned into a contentious issue among photographers.

Putin’s Dream

To craft the Putin’s Dream series, De Keyzer supplied the AI program with his own images from prior works, tailoring it to align with his artistic aesthetic. 

He describes this series as a “reflection on the tragedies of [the Ukraine] conflict precipitated by the aspirations of effectively one individual” and asserts that employing generative AI was a tool to achieve that goal. 

Satisfied with the outcomes, De Keyzer claims the “new visuals — depictions” he has unveiled in Putin’s Dream echo his earlier photographic endeavors, which have frequently delved into propaganda and power structures.

“I attempted to replicate ‘real’ visuals as closely as possible,” he tells ABC News.

“Naturally, it remains synthetic, yet it became feasible to achieve a striking resemblance to nearly realistic images and, more crucially, to incorporate my compositional approach, and commentary [using] irony, humor, doubt, wonder, surrealism … Numerous individuals remark that they can distinctly recognize my style in these visuals, which was the intent.”

De Keyzer emphasizes that he has always been open about his use of AI in creating Putin’s Dream.

However, when he uploaded several images on Instagram to promote his latest publication, the response was severe.

A considerable number of people reprimanded him for sharing “artificial” images, De Keyzer asserts.

“There were an overwhelming amount of negative responses on my Instagram post, roughly 600 in two hours. I was unaccustomed to this. Historically, I received positive reactions to my posts … but this time it was overwhelming … Some claimed they were my biggest supporters previously but not anymore. AI consistently elicits automatic aversion, regardless of the approach or advancements achieved.”

At one point, he feared that the project was a misstep. Nonetheless, he has also garnered support from individuals commending the work, he states.

“Incredible work that illustrates once again how photography can be innovatively approached, without traversing the globe, but rather by exploring another realm, our counterpart, this latent dimension of computational memories that encompass countless layers of accumulated media,” Belgian academic Yves Malicy focused on digital culture remarked in French (this is a translated version) on Facebook.

Is the globe prepared for AI images?

The narrative of photography is punctuated by controversies that involve manipulation, staging, or deception. Nevertheless, photography’s recognition as a depiction of reality persists. As generative AI advances, numerous individuals fear it might unleash an avalanche of misinformation.

When artist Boris Eldagsen stunned the photographic community by securing the Sony World Photography Prize with an AI-generated image, he expressed his aim was to ignite a discussion regarding AI and photography.

“It was a trial to determine if photo contests are ready for [AI]… They are not,” he stated to ABC Radio National.

Contrary to Eldagsen, De Keyzer did not intend to mislead anyone. However, he ultimately removed his Instagram post featuring the images due to backlash, as he claims, against Magnum Photos, the esteemed photographic collective he has belonged to since 1994.

A week subsequent to De Keyzer’s post, Magnum Photos issued a statement concerning AI-generated images.

“[Magnum] honors and cherishes the creative liberty of our photographers,” the statement declared. Yet its archive “will consistently be devoted solely to photographic images captured by humans and that reflect authentic occurrences and narratives, in alignment with Magnum’s heritage and dedication to documentary tradition”.

De Keyzer is not the solitary member of Magnum Photos to incite debate through experimenting with AI image fabrication.

Michael Christopher Brown employed generative AI to create a series of visuals about Cuban refugees. It served as a means to narrate untold stories, he mentioned to PetaPixel.

In a complex reflection on AI, and a “prank” on his photographic community, Jonas Bendiksen utilized software to develop 3D representations of individuals and incorporate them into landscape photographs he captured for a series scrutinizing a Macedonian town that had turned into a notorious center for fake news production. He published a compilation of the visuals titled The Book of Veles, and he utilized AI to generate the text accompanying the book.

“Recognizing that I have fabricated and personally generated false news, I have somehow undermined the credibility of my work,” he stated to Magnum Photos. “Nonetheless, I do hope … that this initiative will enlighten people regarding the future of what lies ahead, and what realms photography and journalism is moving towards.”

The deceiver’s profit

Addressing the Photo Ethics Centre symposium in December, Alex Mahadevan mentioned the erosion of trust instigated by AI-generated visuals which allows individuals to question the authenticity of genuine images or videos is referred to as “the deceiver’s profit”.

Mahadevan, the head of the digital media literacy project MediaWise, cites the Princess Catherine photo incident as an illustration.

After an AI-assisted visual of the princess and her offspring was released, then quickly retracted by

When certain news entities emerged as exceptions, the image triggered rampant conjecture regarding Princess Catherine’s wellness. A clip shared by the princess to inform followers of her condition was subsequently dismissed by many. “Right away, you had individuals across the internet claiming that was not a video of Princess Kate, that it is a deepfake, she is deceased … all of these outrageous conspiracy notions,” Mahadevan states.

This underscores the necessity of openness when it comes to the deployment of generative AI. However, as the panelists at the conference deliberated, aspects like how the application of AI is labeled or recorded in metadata, or when AI supplementation — distinct from generative AI — becomes substantial enough to need disclosure, remain unresolved topics at this juncture.

Savannah Dodd, the founder and director of the Photography Ethics Centre, mentions that there are various ethical considerations, aside from issues of veracity, regarding generative AI technology.

“AI enables creators to generate images of locations they have never personally visited or may have limited knowledge about,” she explains.

Dodd has discussed how biases present in AI image generators, along with a lack of user consultation, can result in the perpetuation of stereotypes.

The decision of which AI generator to utilize should also be thoughtfully weighed, according to Dodd.

“Numerous leading generators aggregate images from throughout the internet, without regard for copyright,” she states.

Last year, photographs of Australian children were discovered within a dataset named LAION-5B, which was employed to train several publicly accessible AI generators that create hyper-realistic images.

In November, a parliamentary investigation into AI issued a report declaring that the entities responsible for generative AI had engaged in “unprecedented theft” from creative professionals in Australia.

The inquiry was presented with a “considerable amount of evidence” indicating that generative AI was already affecting creative industries in Australia.

Dodd asserts that creators working in photography or AI-produced images should evaluate their intentions, the message they wish to express, and the medium they are employing to accomplish this.

“I believe it’s important to take the time to comprehend how an image or collection of images will function in the world, how they will be perceived, and what their possible effects might be,” she adds.

For De Keyzer, the commotion surrounding his application of generative AI is exaggerated. While he believes it’s essential for society to inform itself about AI to prevent potential misuse, he might utilize it again.

“AI is merely another instrument with great potential; why should I have to revert to what I’ve always done?” he states.

“I do appreciate the capability to explore mentally now. I’m getting older, and this could be a means to maintain creativity without the complications and expenses associated with actual travel. Certainly, the real experience is still preferred. Nevertheless, it is becoming increasingly challenging to travel, market the images, and have them published.”