JIANG, Wenzheng; LIANG, Ke; RONG, Xuankun; ZHOU, Jingxuan; ZHONG, Zhengyi; WAN, Guancheng; WANG, Ji. PurMM: Attention-Guided Test-Time Backdoor Purification in Multimodal Large Language Models. Proceedings of the AAAI Conference on Artificial Intelligence, [S. l.], v. 40, n. 42, p. 35562–35570, 2026. DOI: 10.1609/aaai.v40i42.40867. Disponível em: https://ojs.aaai.org/index.php/AAAI/article/view/40867. Acesso em: 14 may. 2026.