H. Hao, “Activation Manipulation Attack: Penetrating and Harmful Jailbreak Attack Against Large Vision-Language Models”, AAAI, vol. 40, no. 42, pp. 35481–35489, Mar. 2026.