Geolocation, Geopolitics and AI in On-Demand Information About Conflicts: a Study of Biases Generative artificial intelligence is increasingly emerging not only as a tool for content creation, but also as a new means of access to knowledge. Previous studies have already shown that, due to their training data, these systems may produce responses affected by different types of bias. This research focuses on potential geographic biases in the responses generated by generative artificial intelligence systems based on large language models (LLMs), taking international conflicts as the object of study. The main objective is to determine whether these systems provide biased responses depending on two variables: the geographical location and the language from which the query is made. The study is based on the hypothesis that such responses do contain biases and that these biases influence users’ perceptions of international conflicts according to their own geographical position. This, in turn, would open up a field of inquiry into the possibilities of the strategic dissemination of particular narratives by political actors through the seeding or control of the training processes of generative AI systems.