Self-Deception: Reverse Penetrating the Semantic Firewall of Large Language Models

TL;DR

This paper introduces a novel 'self-deception' attack method that bypasses semantic defenses in large language models, demonstrating high success rates across multiple languages and scenarios, and providing insights into LLM security vulnerabilities.

Contribution

It proposes the first automatic jailbreak technique using self-deception to penetrate semantic firewalls in LLMs, with extensive multilingual experiments and open-source resources.

Findings

Success rates of 86.2% on GPT-3.5-Turbo and 67% on GPT-4

Generated 2,520 attack payloads across six languages

Effectiveness of self-deception attack demonstrated

Abstract

Large language models (LLMs), such as ChatGPT, have emerged with astonishing capabilities approaching artificial general intelligence. While providing convenience for various societal needs, LLMs have also lowered the cost of generating harmful content. Consequently, LLM developers have deployed semantic-level defenses to recognize and reject prompts that may lead to inappropriate content. Unfortunately, these defenses are not foolproof, and some attackers have crafted "jailbreak" prompts that temporarily hypnotize the LLM into forgetting content defense rules and answering any improper questions. To date, there is no clear explanation of the principles behind these semantic-level attacks and defenses in both industry and academia. This paper investigates the LLM jailbreak problem and proposes an automatic jailbreak method for the first time. We propose the concept of a semantic firewall and provide three technical implementation approaches. Inspired by the attack that penetrates traditional firewalls through reverse tunnels, we introduce a "self-deception" attack that can bypass the semantic firewall by inducing LLM to generate prompts that facilitate jailbreak. We generated a total of 2,520 attack payloads in six languages (English, Russian, French, Spanish, Chinese, and Arabic) across seven virtual scenarios, targeting the three most common types of violations: violence, hate, and pornography. The experiment was conducted on two models, namely the GPT-3.5-Turbo and GPT-4. The success rates on the two models were 86.2% and 67%, while the failure rates were 4.7% and 2.2%, respectively. This highlighted the effectiveness of the proposed attack method. All experimental code and raw data will be released as open-source to inspire future research. We believe that manipulating AI behavior through carefully crafted prompts will become an important research direction in the future.