The Photograph as Evidence – When Can We Trust an Image?

For a long time, it was widely accepted that photography has evidentiary power because what it depicts is not a product of imagination but real, existing phenomena from the external world, captured by the photographer at a specific moment in the past. This belief was not only held by the general public but also formed a fundamental basis for art historical and philosophical discourse about photography. However, numerous examples exist of photographs that have been deliberately altered to misrepresent reality. In the modern era, the rapid spread of digital image-editing tools has made such practices increasingly common. Additionally, the emergence of artificial intelligence has introduced new grounds for skepticism regarding images. To what extent can we trust a photograph? Is such mistrust truly a phenomenon of our time?

For a long time, photography has been regarded as an authentic means of representing reality, capable of recording objects of the external world using light, achieving “the greatest accuracy and faithful preservation of form.”^[1] This capacity endowed photography with a distinctive evidentiary force—one that it alone among the family of images could claim. Capturing a memory, a historical moment, or a person thus also served as proof that the subject depicted was a real, existing entity. In other cases, a photograph attests to the prior existence of its subject, with the image itself acting as tangible evidence.^[2] Beyond art historical and theoretical considerations, this evidentiary role is demonstrated by forensic photography, where the photograph becomes evidence, not metaphorically but in a literal sense. In other contexts, like scientific documentation, photography primarily serves as an observation tool, unveiling phenomena imperceptible to the human eye.^[3] One of the earliest examples of this function is Eadweard Muybridge’s photographic series of a galloping horse. When viewed sequentially, these images not only anticipated the emergence of cinema but also conclusively demonstrated that there is indeed a moment when all four of the horse’s hooves are off the ground.^[4]

However, this required a certain level of technical development in photographic processes. Early light-based image-recording techniques, such as the daguerreotype and the talbotype, required long exposure times; as a result, early urban photographs often depict deserted streets, since moving figures could not be captured by these processes. Placing an engraving and a photograph of the same location side by side allows for a striking comparison. Engravings typically included passersby to animate the scene and enhance its sense of realism, while omitting visual details that were considered unsuitable.^[5] Such images often left out street litter or advertisements pasted on firewalls, thereby presenting a more aesthetic view. In contrast, photographs preserved these incidental details; however, people—if merely passing in front of the camera—failed to appear in the image.^[6]

As photography advanced to a level that allowed for more lifelike representations, it began to completely overshadow engravings, prints, and paintings. During this time, public trust in photographs seemed to grow, even as they became viewed less as works of art and more as small visual embellishments on everyday items. Yet even in this period, cases of photographic falsification are known. To make images more appealing and increase sales, some photographers altered their pictures to enhance their appeal. For example, a photograph of a tornado taken in 1884 by photographer F. N. Robinson has recently come under scrutiny for potential falsification. The image depicts two smaller funnel clouds appearing on either side of a central tornado funnel, with a calm, gently billowing cloud formation above this dramatic atmospheric phenomenon. A scientific research team later pointed out that such tranquil cloud cover cannot form near tornadoes, suggesting that the image was likely produced by combining two separate photographs.^[7]

There are also known examples of photographs that were falsified not for commercial gain but for political purposes. The image-manipulation practices of Soviet propaganda have been extensively studied, revealing numerous photographs from which certain individuals were removed. One of the most famous examples is the retouched image of Nikolai Yezhov. The original photograph shows him walking alongside Stalin in Moscow. However, after losing favor as a key figure during the purges of the late 1930s, Yezhov was executed in early 1940. Following his execution, the photograph also had to be modified: since Yezhov was positioned at the edge of the image, censors were able to erase him with relative ease.^[8] Manipulated photographs appeared even earlier, shortly after Lenin came to power. Individuals considered undesirable or excluded from political life were likewise removed from visual records, which is why Leon Trotsky vanished from several photographs in which he had originally appeared alongside Lenin.^[9]

To this day, cases continue to emerge in which the content of a photograph is called into question, and instances arise in which it later becomes clear that a given image does not depict a real event. Additionally, some works initially believed to be photographs have been found to be generated by artificial intelligence software. A notable controversy arose during the 2023 Sony World Photography Awards when the jury awarded a prize in the Creative category to an AI-generated image. The work was created by Berlin-based photographer Boris Eldagsen, who emphasized repeatedly across multiple platforms after the announcement that, despite any confusion, AI-generated works should not be classified as photographs.^[10] The fully software-generated image sparked widespread debate, with many expressing concern that it is becoming increasingly difficult to distinguish real photographs from computer-generated ones. In this context, the German artist particularly stressed that questions of authenticity should be a central concern and that reliable verification systems are necessary to prevent potential abuses.^[11]

Photography and the representation of reality may thus seem like separate concepts, but there is another perspective to consider. While photographs often serve as evidence in forensic practice, news media, scientific research, or everyday situations, the examples discussed above also demonstrate that photographs also embody the sensibilities and power dynamics of their times, influenced by commercial and political interests. In contrast, images created by artificial intelligence do not represent reality in the same way as photographs do because they do not capture a visual scene from the present. However, AI-generated images can still reflect these societal sensibilities. Despite the growing mistrust around images, photographic processes are likely to maintain their inherent evidentiary nature.