Sandwich Attack

Sandwich attacks exploit vulnerabilities in systems by strategically placing malicious actions between legitimate ones, manipulating the system's response. Research focuses on detecting these attacks, particularly in decentralized finance (using techniques like Support Vector Machines and Graph Attention Networks to analyze transaction data) and large language models (by crafting multi-lingual prompts to elicit harmful outputs). Understanding and mitigating sandwich attacks is crucial for ensuring the security and reliability of both financial technologies and AI systems, driving ongoing efforts to develop more robust detection and prevention mechanisms.