I’m a photojournalist and documentary photographer. I’m also an educator at a School of Journalism. While generative artificial intelligence is capable of positive impacts in a broad range of scientific and creative fields, synthetic visual images made without the physical presence of an eyewitness human reporter using a camera—in particular those made by image generators like DALL-E, Midjourney, and Stable Diffusion—seriously undermine the evidentiary role that camera-led reporting by direct human witness- has played historically in the field of journalism and public interest documentary.

Camera-led reporting and witness are vital to human understanding because when practiced without misleading staging or alteration they are capable of showing us something that AI images (which may look like but are not “photographs,” or “moving-image actuality”) can never show: direct visual evidence made in the specific place and time of unfolding events that engage the credibility of oral testimonies of those living through those events by showing rather than telling.

Still and moving images purporting to be photographic or camera-led eyewitness proof of actual unfolding news events or of non-fiction documentary issues-—whether deployed in newspapers, on TV stations, or on social media—must remain contextualized and easily distinguishable from images which are fabricated by AI. I wholeheartedly support the Statement of Principles by the photojournalism working group on AI, and will adhere to its tenets in my own human-centered non-fiction and journalistic reporting practice and in my teaching.