AI, Deepfakes, and the Crisis of Visual Evidence

This research article examines the crisis of visual evidence emerging from the rapid development of artificial intelligence and synthetic media technologies. It analyzes how AI-generated and AI-altered images—commonly referred to as deepfakes—undermine the traditional epistemic foundations of photography and destabilize long-standing assumptions about the credibility of visual documentation in journalism, human rights reporting, and accountability processes. Drawing on interdisciplinary scholarship in media studies, digital forensics, artificial intelligence ethics, and visual journalism, the article conceptualizes the deepfake phenomenon not merely as a problem of image falsification but as a broader epistemic and institutional disruption. It demonstrates how the erosion of indexical trust affects both fabricated and authentic photographs, enabling strategic denial, accelerating doubt, and weakening the evidentiary function of visual materials in conflict and high-stakes informational environments. The study argues that the crisis of visual evidence cannot be resolved through detection technologies alone. Instead, it calls for a shift from indexical to procedural trust, emphasizing transparency, provenance-by-design, contextual verification, and institutional accountability as the primary foundations for restoring credibility. The article outlines methodological and professional responses aimed at strengthening the resilience of photojournalism under conditions of radical visual uncertainty. The version deposited in Zenodo is presented as a peer-reviewed research article intended to support open academic access, citation, and further interdisciplinary research. The content and analytical structure correspond to the author’s original scholarly contribution and have not been substantively modified.