Political Violence in the Algorithmic Gaze: How Image Search Framed the Russian-Ukrainian War Before and After the 2022 Invasion

Search engines serve as key information intermediaries in today’s high-choice media environment. By selecting and ranking textual and visual content in response to user queries, search engines help individuals navigate the vast amount of information available online and determine what will capture their attention. In this way, search algorithms shape social reality by making certain aspects of specific phenomena, such as race or mass atrocities, more salient than others. Consequently, algorithms that power search engines become crucial elements of the contemporary digital ecosystem, affecting how specific phenomena - including the extreme forms of political violence - are presented to and interpreted by the digital public. To investigate how search algorithms represent present-day political violence, we analyze how two Western (i.e., Google and Bing) and one Russian (i.e., Yandex) search engines visually frame the ongoing Russian-Ukrainian war. Our interest in this case study is due to the fact that this war is the largest conflict in Europe since the 1990s and is also characterized by intense epistemic contestation regarding various aspects of the political violence (e.g., Tyushka, 2023). This epistemic contestation makes the unbiased algorithmic representation of violence more challenging, as search algorithms need to make ontological choices regarding which interpretations or sources to prioritize. The change in the dynamic of the war, which escalated in 2022 following the large-scale Russian invasion and the dramatic increase in the number of deaths and war crimes, poses another challenge that search algorithms have to accommodate. To investigate how search algorithms tackle these challenges, we use a unique longitudinal dataset on the search engine representation of the Russian-Ukrainian war in 2021-2025. To collect the dataset, we employed a virtual agent-based approach (Haim et al., 2017) and regularly audited the performance of Google, Yandex, and Bing for the query “war in Ukraine” before and after the escalation of the war in 2022. Google and Bing are the two most commonly used search engines in the Global North; by contrast, Yandex is the major search engine in Russia and is also particularly prone to the influence of the Kremlin. To analyze image search results from these engines, we employ qualitative content analysis and examine the consistency of the salience of specific aspects of the war and actors over time, together with differences in saliency across search engines. Specifically, we are interested in whether the algorithms are subject to representation bias, for instance, by adopting the male gaze on the war or systematically underrepresenting specific aspects of violence. We also examine how the representation of violence evolves over time and whether Yandex promotes a more pro-Kremlin portrayal of the war compared to the other two Western search engines.