LLM-PEA: Leveraging Large Language Models Against Phishing Email Attacks

2025-12-10

AI Summary

Large language models (LLMs) deployed in email security systems are vulnerable to coordinated multi-vector attacks—including prompt injection, text refinement, and cross-lingual adversarial perturbations—yet no native LLM-based detection framework exists for phishing emails. Method: We propose the first LLM-native multi-vector phishing detection framework, integrating GPT-4o, Claude Sonnet 4, and Grok-3 with systematic prompt engineering, cross-lingual semantic alignment, and rigorous adversarial testing. Contribution/Results: Our framework achieves >90% phishing detection accuracy and, for the first time, empirically characterizes the robustness boundaries and composable vulnerabilities of LLMs under realistic phishing scenarios. Experiments reveal critical failure modes in existing LLM-based detectors under coordinated prompt injection and multilingual attacks. The framework establishes a reproducible evaluation benchmark and a principled defense paradigm for secure LLM deployment in email security.

Technology Category

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

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Email phishing is one of the most prevalent and globally consequential vectors of cyber intrusion. As systems increasingly deploy Large Language Models (LLMs) applications, these systems face evolving phishing email threats that exploit their fundamental architectures. Current LLMs require substantial hardening before deployment in email security systems, particularly against coordinated multi-vector attacks that exploit architectural vulnerabilities. This paper proposes LLMPEA, an LLM-based framework to detect phishing email attacks across multiple attack vectors, including prompt injection, text refinement, and multilingual attacks. We evaluate three frontier LLMs (e.g., GPT-4o, Claude Sonnet 4, and Grok-3) and comprehensive prompting design to assess their feasibility, robustness, and limitations against phishing email attacks. Our empirical analysis reveals that LLMs can detect the phishing email over 90% accuracy while we also highlight that LLM-based phishing email detection systems could be exploited by adversarial attack, prompt injection, and multilingual attacks. Our findings provide critical insights for LLM-based phishing detection in real-world settings where attackers exploit multiple vulnerabilities in combination.

Problem

• Detects phishing emails using LLMs across multiple attack vectors.
• Evaluates LLM robustness against adversarial and multilingual phishing threats.
• Addresses vulnerabilities in LLM-based email security systems.

Innovation

• LLM-based framework for multi-vector phishing detection
• Evaluates three frontier LLMs with comprehensive prompting design
• Empirical analysis reveals over 90% detection accuracy