The initial response to an AI-related security incident should focus on containment. Isolating the AI system helps prevent further unauthorized access, data leakage, model misuse, or spread to connected systems while preserving evidence for investigation.