MAI-2025-0017

Large Language Models LLMs are susceptible to a sophisticated adversarial attack known as Alphabet Index Mapping AIM. This attack demonstrates a high efficacy in circumventing safety filters, commonly referred to as "jailbreaking." AIM operates by encoding prompts through the conversion of...

Alphabet Index Mapping: Jailbreaking LLMs through Semantic Dissimilarity

Large Language Models LLMs have demonstrated remarkable capabilities, yet their susceptibility to adversarial attacks, particularly jailbreaking, poses significant safety and ethical concerns. While numerous jailbreak methods exist, many suffer from computational expense, high token usage, or...