Defend LLMs Through Self-Consciousness

This paper introduces a novel self-consciousness defense mechanism for Large Language Models (LLMs) to combat prompt injection attacks. Unlike traditional approaches that rely on external classifiers, our method leverages the LLM's inherent reasoning capabilities to perform self-protection. We propose a framework that incorporates Meta-Cognitive and Arbitration Modules, enabling LLMs to evaluate and regulate their own outputs autonomously. Our approach is evaluated on seven state-of-the-art LLMs using two datasets: AdvBench and Prompt-Injection-Mixed-Techniques-2024. Experiment results demonstrate significant improvements in defense success rates across models and datasets, with some achieving perfect and near-perfect defense in Enhanced Mode. We also analyze the trade-off between defense success rate improvement and computational overhead. This self-consciousness method offers a lightweight, cost-effective solution for enhancing LLM ethics, particularly beneficial for GenAI use cases across various platforms.

Key Contributions

Self-consciousness framework that leverages LLMs' inherent reasoning to detect and block prompt injection without external classifiers
Meta-Cognitive and Arbitration Modules enabling autonomous output evaluation and regulation
Evaluation across seven state-of-the-art LLMs on two prompt injection datasets showing near-perfect defense rates in Enhanced Mode with analysis of computational trade-offs