M. Aktas

Unmasking Toxicity and Vulnerabilities in Large Vision-Language Models

This study systematically examines the vulnerabilities of open and closed-weight LVLMs, including LLaVA, InstructBLIP, Fuyu, Qwen, DeepSeek, Gemini, GPT, and Grok, using adversarial prompting strategies informed by social theories to simulate real-world social manipulation tactics.