Code for the 2025 ACL publication "ConInstruction: Universal Jailbreaking of Multimodal Large Language Models via Non-Textual Modalities" - View it on GitHub