Artificial intelligence and the image war: does exposing AI-generated images as fake limit the images’ influence on perceptions about wars and conflicts?

The emergence of artificial intelligence (AI) has reshaped how visual narratives influence public perception during conflicts, challenging long-held assumptions about wartime image credibility and emotional impact. While research has explored AI’s role in disinformation, its influence on wartime narrative formation remains under-studied. Through an experimental survey conducted in the United States in August 2024 (N = 1812) using the 2023–2024 Israel-Hamas conflict as a case study, participants were exposed to four AI-generated images presenting the Palestinian narrative. The study measured attitudinal changes toward conflict actors before and after image exposure, with half the participants being subsequently informed of the images’ artificial nature. Initial findings revealed that AI-generated images effectively influenced attitudes regardless of disclosure status, with both groups showing increased support for the Palestinian narrative. However, when informed about the artificial nature of the images, participants demonstrated a significant attitudinal shift compared to those who were kept unaware, exhibiting increased support for Israel rather than the Palestinians whose narrative the images portrayed. This ‘backfire effect’ persisted even when controlling for pre-existing attitudes and demographic factors. The findings extend our understanding of both visual framing and algorithm aversion theories in conflict settings, suggesting that, in warfare communication, algorithm aversion manifests more powerfully than previously theorized. These results have significant implications for political actors’ strategic communication choices and media organizations’ disclosure policies regarding AI-generated content in conflict coverage.