‘A photo captures illumination on a sensor or film. It serves as a record of a tangible moment in time.’
‘A photo captures illumination on a sensor or film. It serves as a record of a tangible moment in time.’


We enjoy pondering about how the concept of “real” photography is interpreted here at The Verge now that generative AI is widespread, and the World Press Photo contest might provide clarity. This renowned accolade honors the pinnacle of photojournalism, where depicting reality holds great importance.
The recipient of the 2026 award — “Separated by ICE,” taken by photojournalist Carol Guzy — was revealed yesterday. The distressing image portrays children holding onto their father following an immigration hearing. The image needed to conform to specific regulations concerning AI tool usage to qualify for the contest, with the independent nonprofit organization that oversees the award asserting that “AI-generated visuals do not count as photography.”
“A photo captures illumination on a sensor or film,” the organization articulated in its entry guidelines. “It serves as a record of a tangible moment in time.” The organization also specifies the precise tools and equipment allowed to qualify:
Every photograph submitted to the contest must be produced with a camera. No synthetic or artificially generated images will be accepted, and no use of generative fill is permitted in post-processing. Any use of these tools will automatically disqualify the submission from the contest.
Photos taken with smartphones are allowed only if captured in the standard shooting mode; entries produced using HDR, portrait mode, creative lighting effects, or panorama mode are ineligible.
The use of smart tools or AI-driven enhancement tools is allowed within the contest rules, as long as these tools do not result in significant alterations to the image overall, introduce new information to the image, or eliminate information from the image that was recorded by the camera.
Tools that immediately violate the contest regulations and are prohibited include all AI-powered enlarging tools like Adobe Super Resolution and Topaz Photo AI. These are based on generative AI models that add new information to enhance and enlarge images.
There remain some ambiguous areas, such as using AI tools for reducing noise, automatic color/contrast adjustments, and object selection, which are allowed as long as they’re not applied excessively and don’t “alter, add or remove details from the photograph that was captured by the camera,” per World Press Photo. The organization also provides a link to a page that outlines what constitutes photo manipulation according to its rules.
It features one of the most thorough sets of requirements we’ve encountered for outlining what qualifies as genuine photography, and could act as a standard for how we, the general public, contemplate this subject in the future.