Logic Jailbreak: Efficiently Unlocking LLM Safety Restrictions Through Formal Logical Expression

TL;DR

This paper introduces LogiBreak, a logical expression-based method to bypass LLM safety restrictions by translating harmful prompts into formal logic, exposing vulnerabilities in current safety mechanisms.

Contribution

It presents a novel black-box jailbreak technique using logical translation to effectively evade LLM safety systems across multiple languages.

Findings

LogiBreak successfully bypasses safety constraints in multilingual tests.

The method exploits distributional gaps between natural language prompts and logical expressions.

Effective across various evaluation settings and linguistic contexts.

Abstract

Despite substantial advancements in aligning large language models (LLMs) with human values, current safety mechanisms remain susceptible to jailbreak attacks. We hypothesize that this vulnerability stems from distributional discrepancies between alignment-oriented prompts and malicious prompts. To investigate this, we introduce LogiBreak, a novel and universal black-box jailbreak method that leverages logical expression translation to circumvent LLM safety systems. By converting harmful natural language prompts into formal logical expressions, LogiBreak exploits the distributional gap between alignment data and logic-based inputs, preserving the underlying semantic intent and readability while evading safety constraints. We evaluate LogiBreak on a multilingual jailbreak dataset spanning three languages, demonstrating its effectiveness across various evaluation settings and linguistic contexts.