Pictures, but Did It Happen? An Inquiry on the Epistemic Value of News Photography after AI Image Generation
LESS INFO

ABSTRACT

Photojournalism sanctifies the belief that one photograph can stop the war; a political aim of “civic witnessing” grounded in a photographic ontology of indexicality. This substantiates an epistemology of photography as an ‘objective’ view. Artificial Intelligence disrupts this epistemology and transforms the objective view into a liability for civic witnessing. This means of claiming epistemic value relies on an ontological relationship elevated to the level of myth: objectivity necessarily skips over the labor of “forging” a photograph into an object capable of making knowledge claims. This research explores this labor, asking: What is the epistemic value of the news photograph today? I conducted ethnographic fieldwork at those sites where photographic meaning is negotiated. I detail three moments of knowledge forging, and the epistemic norms which underpin (and are challenged by) them, within: photojournalistic practices, photorealistic AI-generated images, and photographs accused of AI. These cases reveal photographic meaning is negotiated in circulation, by traversing epistemic infrastructures, those networked sites where a photograph can gain and be evacuated of meaning. This research uncovers how epistemic norms relativize across online publics, intersecting with claims to objectivity to contradictory ends. What alternatives might there be? I argue, following Donna Haraway, for an epistemology which furnishes news photography’s epistemic value as “situated knowledges.” In practice, this requires attention to the pathways a photograph must traverse and the moments when it fails to do so, the negotiations of permissible sights which precede public eyes, in short, the embodied entanglements and situated views behind photographic meaning.

FIELDWORK

I defined my fieldsite as the places where image-knowledge was negotiated, broadly construed. This included visual journalists that create news images, human rights investigators who work with OSINT to verify images and video and political subcultures on social media which engage with and circulate images. My interest was in how these spaces operate together, often to contradictory ends, to create a wider sense of “what happened” at any given news event. 

Conversations with photojournalists were guided by several questions: How does AI impact public trust in visual journalism? How does this compare to earlier discussions on, for example, photoshop, filtering and image-editing? What strategies do visual journalists employ to secure public trust? Where do these strategies no longer work today? My participants regularly shared that accusations of AI have been used to discredit their work. We also discussed how social media virality fosters an environment of doubt and, crucially, how an institutional reliance on objectivity to ground knowledge claims fails to address new forms of doubt facing photojournalistic practice today.

Human rights investigators operate further down the line of the doubt problem. I spoke both with professional OSINT investigators and participated in a visual investigation with the University of Cambridge’s Digital Verification Corps, where we verified and wrote a report on the IDF’s use of white phosphorus airstrikes in southern Lebanon. Here, it was interesting to see how investigators handled the tension between subjective witness accounts and the need for objective evidence. I saw how investigators transformed, through satellite imagery and visual documentation from several angles, social media accounts into evidence for human rights reportage. 

Beyond these traditional fieldwork components, I spent several months embedded with political subcultures. I tracked how users engaged with news images, whether photographs or AI-generated, what conditions allowed these images to be seen as true, and how users operationalized AI-generated images to make customizable knowledge claims. This part of the project looked particularly at AI-generated images purporting to show U.S. soldiers in Iran. 

UPDATES

The project was completed in June 2026 for my MPhil in Digital Humanities, where it was awarded a Starred Distinction (80). I’m currently transforming the work into a journal article. 

AI generated image, X
Roger Fenton, Valley of the Shadow of Death