During the Iran war in early 2026, users increasingly turned to xAI’s Grok chatbot on X to verify viral videos and images, but the tool frequently confirmed AI-generated or unrelated footage as authentic. Grok misidentified a real video from Tehran as being from Italy, described genuine footage of Israeli strikes near Tehran as a 2017 California wildfire, and misattributed footage of a strike on a school in Iran to a 2021 attack in Kabul, contradicting verification later performed by the New York Times, Reuters, and Bellingcat. Grok’s answers about the same footage were also inconsistent between users, sometimes labeling identical content as both real and fabricated.