Every photograph, at its most fundamental, is a record of light that actually existed. Photons bounced off real surfaces, passed through a real lens, and struck a surface that captured their pattern. For nearly two centuries, this has been the unspoken contract between the photographer and the viewer: what you are looking at happened. Someone was there. This is evidence.
That contract is now under renegotiation.
We have arrived at a peculiar moment in visual culture. Generative AI can now produce images that are, to most eyes, indistinguishable from photographs. A portrait with convincing skin texture and catch-lights in the eyes. A landscape with atmospheric perspective and the right quality of shadow for late afternoon. A photojournalistic scene complete with motion blur and the slightly imperfect framing that signals authenticity. These images depict nothing that ever existed. No light was gathered. No one stood in a particular place at a particular time and chose to press the shutter.
For commercial photography, illustration, and certain branches of creative image-making, this is a disruption worth taking seriously on its own terms. But the implications run deeper than displaced stock photographers and altered business models. What generative AI fundamentally challenges is photography’s century-old claim to be a document of reality, and that challenge matters far beyond the photography industry itself.
Consider how thoroughly we have built our shared understanding of the world on photographic evidence. Court proceedings, insurance claims, news reports, scientific research, human rights documentation, historical archives. The photograph has served as a kind of visual oath, a declaration that this happened, here, then. We have always known, of course, that photographs could be manipulated. Darkroom techniques, Photoshop, careful staging. But the manipulation required effort and left traces. The baseline assumption remained intact: a photograph, until proven otherwise, represented something real.
Generative AI inverts that assumption. When photorealistic images can be conjured from text prompts in seconds, the burden shifts. The question is no longer “has this image been altered?” but rather “was this image ever a photograph at all?” This is not a subtle distinction. It represents a fundamental change in how visual evidence functions in society.
It would be easy, at this point, to succumb to a kind of visual apocalypticism. The end of photographic truth. The death of the image as evidence. But history suggests that the reality will be both less dramatic and more interesting than the headlines imply.
Photography has weathered several existential crises before, and each time the medium has emerged not destroyed but clarified. When photography was first invented, painters worried it would make their craft obsolete. What actually happened was more nuanced: painting was liberated from the obligation to faithfully reproduce appearances, giving rise to Impressionism and eventually abstraction. Photography did not kill painting. It freed painting to become more fully itself.
The arrival of digital photography in the 1990s provoked similar anxieties. Without the negative, sceptics argued, the photograph would lose its evidentiary weight. A digital file could be altered without trace. And while those concerns were not unfounded, what actually evolved was a more sophisticated understanding of digital provenance, metadata, and authentication. The medium adapted. Institutions developed new frameworks for verifying digital images. Photojournalism established stricter ethical codes. The tools changed; the underlying commitment to truthful documentation persisted.
This pattern, where a new technology appears to threaten a medium’s core purpose, only to ultimately refine our understanding of what that purpose actually is, repeats throughout creative and technical history. It would be unwise to assume this iteration will be different, even as we acknowledge that the scale and speed of generative AI present genuinely novel challenges.
If we accept that AI-generated images will become increasingly pervasive, and there seems little reason to doubt this, then a more productive question emerges: what does photography offer that generation cannot replicate?
The answer, perhaps surprisingly, has less to do with the final image than with the process of its creation.
A photograph is the product of physical presence. Someone stood in a specific location, at a specific moment, and made a series of decisions about what to include in the frame and what to leave out. They responded to light as it actually fell, to events as they actually unfolded. The resulting image carries what we might call situated knowledge, an understanding that comes only from being there. This is not a quality that can be prompted into existence.
Documentary photography, in particular, derives its power from this irreducible connection to lived reality. When we look at Don McCullin’s images from conflict zones, or Dorothea Lange’s Depression-era portraits, or Sebastião Salgado’s documentation of human labour, we are not simply looking at compelling compositions. We are looking at evidence of encounter. A human being was present with another human being, in conditions that were often difficult or dangerous, and chose to bear witness. The photograph is the residue of that act of witnessing, and no amount of computational sophistication can fabricate it.
This extends beyond the dramatic examples. Even in quieter documentary work, the act of spending time in a place, learning its rhythms, earning the trust of its people, and waiting for the moment when light, gesture, and meaning converge represents a form of understanding that is fundamentally embodied. I recall mornings spent in the Welsh valleys during my own early photographic education, where the landscape taught patience in a way no instruction could. The mist would sit in the valley bottoms, the light would change by the minute, and you learned, gradually, that the image you had imagined was rarely the image the place wanted to give you. That negotiation between intention and reality, that willingness to be changed by what you encounter, sits at the heart of what makes documentary photography more than mere image-making.
Against this backdrop, the growing resurgence of analogue photography begins to make a different kind of sense. On the surface, the return to film, darkroom processing, and historical techniques like wet plate collodion or cyanotype might appear to be nostalgia, a retreat to the aesthetic comforts of a pre-digital age. But something more significant is happening.
Traditional photographic processes create what we might think of as an unbroken chain of physical evidence. In a wet plate collodion image, for instance, the glass plate that was present at the moment of exposure is the final image. There is no file, no intermediate digital step, no possibility of generative insertion. The image is, in the most literal sense, a physical artefact of the light that existed at a particular moment. It carries its own authentication in its materiality.
Film photography, while not quite as dramatically singular, operates on a similar principle. The negative is a physical record. The grain is the actual structure of silver halide crystals responding to photons. The slight unpredictability of chemical processes, the way a particular emulsion renders colour, the tonal characteristics of a specific developer, these are not filters or presets. They are the fingerprints of a real, physical process acting on real, captured light.
This matters now in ways it did not a decade ago. When every digital image exists under a cloud of suspicion about its origins, the physicality of analogue photography offers something remarkably valuable: verifiable authenticity. Not because analogue images cannot be staged or manipulated, they absolutely can, but because the analogue process creates a tangible, examinable chain of evidence from the moment of exposure to the final print. In an age of synthetic images, this material connection to reality becomes not a quaint limitation but a profound asset.
There is also something to be said for what analogue constraints do to the photographer’s practice. Working with a limited number of exposures on a roll of film changes how you see. It demands a decisiveness and intentionality that infinite digital storage quietly erodes. Each frame costs something, not merely in money but in opportunity. This economy of means tends to produce a more considered, more deliberate form of seeing. It is not that digital photography cannot be practised with equal intentionality, of course it can, but the constraints of analogue make that intentionality structural rather than merely aspirational.
For documentary photography and photojournalism, these questions of authenticity and evidence are not philosophical abstractions. They are practical and urgent.
Several news organisations and photojournalistic bodies are already grappling with how to authenticate images in a world saturated with AI-generated content. The Coalition for Content Provenance and Authenticity (C2PA) and similar initiatives are developing metadata standards that track an image’s creation and editing history. These are important technical responses, but they address the problem at the level of verification rather than creation.
The more fundamental response may be a renewed emphasis on the photographic process itself as a form of evidence. When a photographer can demonstrate not just the final image but the contact sheets, the unprocessed negatives, or the original RAW files with unbroken metadata chains, they are providing something that generative AI cannot: proof of presence. The negative, in this sense, becomes more than a technical artefact. It becomes a kind of witness statement.
This is where traditional and analogue practices offer something that transcends aesthetics. A large format photographer working with sheet film produces a unique, physical negative for every exposure. A wet plate practitioner creates an unreproducible artefact. Even a photographer working with 35mm film generates a physical record that can be examined, verified, and authenticated in ways that a digital file, however carefully managed, cannot entirely match.
We should be careful not to overclaim here. The answer to generative AI is not a wholesale retreat to nineteenth-century processes. Digital photography remains extraordinary in its capabilities, its accessibility, and its capacity for documentation at scale. The most thoughtful response will almost certainly involve both analogue authenticity and robust digital provenance systems working in concert rather than in opposition.
What generative AI is doing to photography is not, in the end, destroying it. It is clarifying it. By creating a category of images that look like photographs but are not, AI is compelling us to articulate what we actually value in a photograph, and it turns out that what we value is not simply the way it looks.
We value the presence of the photographer. We value the moment of encounter. We value the physical connection between light, lens, and recording surface. We value the constraints that force genuine creative decisions. We value the evidence of having been there.
These are qualities that have always been central to photography, but we did not need to name them when they could be taken for granted. Now that they can no longer be assumed, we are forced to recognise them, and in doing so, we may find that we understand the medium more deeply than we did before.
The return to traditional photographic techniques, far from being a backwards step, may represent the clearest articulation of what photography has always been at its core: a practice of witness. A way of being present in the world, attentive to its light, and committed to the faithful recording of what was actually there. In a culture increasingly saturated with images that depict nothing real, that commitment to the real may prove to be photography’s most radical and valuable quality.
The light was there. Someone gathered it. That, perhaps, is enough.
