Efficient Detection of Toxic Prompts in Large Language Models

TL;DR

This paper introduces ToxicDetector, a lightweight greybox approach that efficiently detects toxic prompts in large language models with high accuracy and low false positives, suitable for real-time applications.

Contribution

The paper presents ToxicDetector, a novel method combining toxic concept prompts, embedding features, and an MLP classifier to improve detection of toxic prompts in LLMs.

Findings

Achieves 96.39% accuracy in toxic prompt detection.

Low false positive rate of 2.00%.

Processing time of 0.078 seconds per prompt.

Abstract

Large language models (LLMs) like ChatGPT and Gemini have significantly advanced natural language processing, enabling various applications such as chatbots and automated content generation. However, these models can be exploited by malicious individuals who craft toxic prompts to elicit harmful or unethical responses. These individuals often employ jailbreaking techniques to bypass safety mechanisms, highlighting the need for robust toxic prompt detection methods. Existing detection techniques, both blackbox and whitebox, face challenges related to the diversity of toxic prompts, scalability, and computational efficiency. In response, we propose ToxicDetector, a lightweight greybox method designed to efficiently detect toxic prompts in LLMs. ToxicDetector leverages LLMs to create toxic concept prompts, uses embedding vectors to form feature vectors, and employs a Multi-Layer Perceptron (MLP) classifier for prompt classification. Our evaluation on various versions of the LLama models, Gemma-2, and multiple datasets demonstrates that ToxicDetector achieves a high accuracy of 96.39\% and a low false positive rate of 2.00\%, outperforming state-of-the-art methods. Additionally, ToxicDetector's processing time of 0.0780 seconds per prompt makes it highly suitable for real-time applications. ToxicDetector achieves high accuracy, efficiency, and scalability, making it a practical method for toxic prompt detection in LLMs.