This paper examines how generative AI LLM architectures can mitigate the gap between information velocity and verification capacity in crisis-driven information environments.Drawing on the case of a newsroom-deployed fact-checking system developed in wartime Ukraine, we show how verification evolved from manual workflows to a hybrid architecture combining claim classification, evidence retrieval, synthesis, and human oversight.We conceptualize this as a response to a structural velocitycapacity
