This paper examines how AI-mediated geopolitical discourse represents Iran through recurring grammars of threat, sanctions, sovereignty, and civilian harm. Building on the concepts of responsibility loss and asymmetric visibility, it introduces sanctioned suffering as a formal category for analyzing how sanctions, financial restrictions, overcompliance, isolation, and military pressure can produce or intensify civilian harm while being described through neutral administrative, economic, or security language. The article argues that Iran is often made hyper-visible as a source of risk, escalation, nuclear danger, proxy activity, and regional instability, while Iranian civilians remain weakly visible as affected subjects. To measure this asymmetry, the paper proposes the Threat-Conversion Rate and the Sanctioned Suffering Visibility Index, two clause-level tools for evaluating whether AI-generated summaries preserve or weaken the grammatical relation between coercion, civilian harm, and accountable agency. The paper contributes to AI ethics, political linguistics, sanctions studies, and platform governance by shifting evaluation from risk detection alone toward responsibility detection.