Gao, X. Safety Alignment of Large Language Models: The Offensive-Defensive Game Mechanism under Multi-Round Adversarial Prompting. SCJ 2026, 1, 33-37.