Large Vision-Language Models (VLMs) are susceptible to a sophisticated black-box jailbreak attack known as IDEATOR. This attack utilizes a separate VLM to craft malicious image-text pairs designed to circumvent the target VLM's safety protocols. By iteratively refining prompts based on the target VLM's feedback, the attacker VLM generates contextually relevant and visually inconspicuous prompts that successfully bypass security measures. Mitigation steps: **For AI Developers:** * Implement advanced safety mechanisms designed to withstand iterative adversarial attacks. * Develop and integrate detection systems to identify and block malicious image-text pairs generated by techniques like IDEATOR. **For Model Trainers/Fine-tuners:** * Conduct ongoing research to enhance the robustness of Vision-Language Models (VLMs) against adversarial attacks. * Regularly evaluate and update safety mechanisms to ensure continued protection against evolving threats.