Sentra-Guard: A Multilingual Human-AI Framework for Real-Time Defense Against Adversarial LLM Jailbreaks

2025-10-26

AI Summary

Jailbreaking and prompt injection attacks pose critical security threats to large language models (LLMs). Method: This paper proposes a real-time, modular, multilingual defense system comprising a classifier-retriever fusion module—integrating SBERT-based semantic retrieval (via FAISS indexing) with a fine-tuned Transformer classifier—augmented by a language-agnostic preprocessing layer, a dual-label knowledge base, and a human-in-the-loop feedback mechanism for continual learning. Contribution/Results: The system introduces the first context-aware dynamic risk scoring mechanism and achieves language-unbiased protection across 100+ languages. It supports transparent, tunable deployment across multiple LLMs. Experimental evaluation demonstrates exceptional performance: 99.96% detection accuracy, AUC = 1.00, F1 = 1.00, and only 0.004% attack success rate—significantly outperforming baselines including LlamaGuard-2 and OpenAI Moderation.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Safety and Robustness
• Computer Vision: Large Vision Models

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

This paper presents a real-time modular defense system named Sentra-Guard. The system detects and mitigates jailbreak and prompt injection attacks targeting large language models (LLMs). The framework uses a hybrid architecture with FAISS-indexed SBERT embedding representations that capture the semantic meaning of prompts, combined with fine-tuned transformer classifiers, which are machine learning models specialized for distinguishing between benign and adversarial language inputs. It identifies adversarial prompts in both direct and obfuscated attack vectors. A core innovation is the classifier-retriever fusion module, which dynamically computes context-aware risk scores that estimate how likely a prompt is to be adversarial based on its content and context. The framework ensures multilingual resilience with a language-agnostic preprocessing layer. This component automatically translates non-English prompts into English for semantic evaluation, enabling consistent detection across over 100 languages. The system includes a HITL feedback loop, where decisions made by the automated system are reviewed by human experts for continual learning and rapid adaptation under adversarial pressure. Sentra-Guard maintains an evolving dual-labeled knowledge base of benign and malicious prompts, enhancing detection reliability and reducing false positives. Evaluation results show a 99.96% detection rate (AUC = 1.00, F1 = 1.00) and an attack success rate (ASR) of only 0.004%. This outperforms leading baselines such as LlamaGuard-2 (1.3%) and OpenAI Moderation (3.7%). Unlike black-box approaches, Sentra-Guard is transparent, fine-tunable, and compatible with diverse LLM backends. Its modular design supports scalable deployment in both commercial and open-source environments. The system establishes a new state-of-the-art in adversarial LLM defense.

Problem

• Detects and mitigates adversarial jailbreak attacks on LLMs
• Identifies malicious prompts across multiple languages automatically
• Reduces false positives with human-in-the-loop feedback system

Innovation

• Hybrid FAISS-SBERT architecture for semantic prompt analysis
• Classifier-retriever fusion module for context-aware risk scoring
• Multilingual preprocessing with human-in-the-loop feedback system