Clouding the Mirror: Stealthy Prompt Injection Attacks Targeting LLM-based Phishing Detection

• URL: http://arxiv.org/abs/2602.05484v1
• Date: Thu, 05 Feb 2026 09:44:20 GMT
• Title: Clouding the Mirror: Stealthy Prompt Injection Attacks Targeting LLM-based Phishing Detection
• Authors: Takashi Koide, Hiroki Nakano, Daiki Chiba,
• Abstract summary: Large language models (LLMs) are vulnerable to prompt injection (PI)<n>This paper presents the first comprehensive evaluation of PI against multimodal LLM-based phishing detection.<n>We propose InjectDefuser, a defense framework that combines prompt hardening, allowlist-based retrieval augmentation, and output validation.
• Score: 0.7391823486666542
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Phishing sites continue to grow in volume and sophistication. Recent work leverages large language models (LLMs) to analyze URLs, HTML, and rendered content to decide whether a website is a phishing site. While these approaches are promising, LLMs are inherently vulnerable to prompt injection (PI). Because attackers can fully control various elements of phishing sites, this creates the potential for PI that exploits the perceptual asymmetry between LLMs and humans: instructions imperceptible to end users can still be parsed by the LLM and can stealthily manipulate its judgment. The specific risks of PI in phishing detection and effective mitigation strategies remain largely unexplored. This paper presents the first comprehensive evaluation of PI against multimodal LLM-based phishing detection. We introduce a two-dimensional taxonomy, defined by Attack Techniques and Attack Surfaces, that captures realistic PI strategies. Using this taxonomy, we implement diverse attacks and empirically study several representative LLM-based detection systems. The results show that phishing detection with state-of-the-art models such as GPT-5 remains vulnerable to PI. We then propose InjectDefuser, a defense framework that combines prompt hardening, allowlist-based retrieval augmentation, and output validation. Across multiple models, InjectDefuser significantly reduces attack success rates. Our findings clarify the PI risk landscape and offer practical defenses that improve the reliability of next-generation phishing countermeasures.

Related papers

• The Trojan Knowledge: Bypassing Commercial LLM Guardrails via Harmless Prompt Weaving and Adaptive Tree Search [58.8834056209347]
  Large language models (LLMs) remain vulnerable to jailbreak attacks that bypass safety guardrails to elicit harmful outputs.<n>We introduce the Correlated Knowledge Attack Agent (CKA-Agent), a dynamic framework that reframes jailbreaking as an adaptive, tree-structured exploration of the target model's knowledge base.
  arXiv  Detail & Related papers  (2025-12-01T07:05:23Z)
• SoK: Large Language Model-Generated Textual Phishing Campaigns End-to-End Analysis of Generation, Characteristics, and Detection [3.7549350220109274]
  Large language models (LLMs) enable Phishing-as-a-Service'' attacks at scale within minutes.<n>Despite the growing research into LLM-facilitated phishing attacks, consolidated systematic research on the phishing attack life cycle remains scarce.<n>We present the first systematization of knowledge (SoK) on LLM-generated phishing, offering an end-to-end analysis that spans generation techniques, attack features, and mitigation strategies.
  arXiv  Detail & Related papers  (2025-08-29T09:39:46Z)
• BlindGuard: Safeguarding LLM-based Multi-Agent Systems under Unknown Attacks [58.959622170433725]
  BlindGuard is an unsupervised defense method that learns without requiring any attack-specific labels or prior knowledge of malicious behaviors.<n>We show that BlindGuard effectively detects diverse attack types (i.e., prompt injection, memory poisoning, and tool attack) across multi-agent systems.
  arXiv  Detail & Related papers  (2025-08-11T16:04:47Z)
• Adaptive Linguistic Prompting (ALP) Enhances Phishing Webpage Detection in Multimodal Large Language Models [2.859540477694109]
  This study explores few-shot Adaptive Linguistic Prompting (ALP) in detecting phishing webpages.<n>ALP is a structured semantic reasoning method that guides large language models (LLMs) to analyze textual deception.<n>Our experiments demonstrate that ALP significantly enhances phishing detection accuracy.
  arXiv  Detail & Related papers  (2025-06-29T01:26:25Z)
• MultiPhishGuard: An LLM-based Multi-Agent System for Phishing Email Detection [3.187381965457262]
  MultiPhishGuard is a dynamic multi-agent detection system that synergizes specialized expertise with adversarial-aware reinforcement learning.<n>Our framework employs five cooperative agents with automatically adjusted decision weights powered by a Proximal Policy Optimization reinforcement learning algorithm.<n>Experiments demonstrate that MultiPhishGuard achieves high accuracy (97.89%) with low false positive (2.73%) and false negative rates (0.20%)
  arXiv  Detail & Related papers  (2025-05-26T23:27:15Z)
• Defending against Indirect Prompt Injection by Instruction Detection [109.30156975159561]
  InstructDetector is a novel detection-based approach that leverages the behavioral states of LLMs to identify potential IPI attacks.<n>InstructDetector achieves a detection accuracy of 99.60% in the in-domain setting and 96.90% in the out-of-domain setting, and reduces the attack success rate to just 0.03% on the BIPIA benchmark.
  arXiv  Detail & Related papers  (2025-05-08T13:04:45Z)
• Next-Generation Phishing: How LLM Agents Empower Cyber Attackers [10.067883724547182]
  The escalating threat of phishing emails has become increasingly sophisticated with the rise of Large Language Models (LLMs) As attackers exploit LLMs to craft more convincing and evasive phishing emails, it is crucial to assess the resilience of current phishing defenses. We conduct a comprehensive evaluation of traditional phishing detectors, such as Gmail Spam Filter, Apache SpamAssassin, and Proofpoint, as well as machine learning models like SVM, Logistic Regression, and Naive Bayes. Our results reveal notable declines in detection accuracy for rephrased emails across all detectors, highlighting critical weaknesses in current phishing defenses.
  arXiv  Detail & Related papers  (2024-11-21T06:20:29Z)
• PEEK: Phishing Evolution Framework for Phishing Generation and Evolving Pattern Analysis using Large Language Models [10.455333111937598]
  Phishing remains a pervasive cyber threat, as attackers craft deceptive emails to lure victims into revealing sensitive information.<n>Deep learning has become a key component in defending against phishing attacks, but these approaches face critical limitations.<n>We propose the first Phishing Evolution FramEworK (PEEK) for augmenting phishing email datasets with respect to quality and diversity.
  arXiv  Detail & Related papers  (2024-11-18T09:03:51Z)
• From ML to LLM: Evaluating the Robustness of Phishing Webpage Detection Models against Adversarial Attacks [0.8050163120218178]
  Phishing attacks attempt to deceive users into stealing sensitive information, posing a significant cybersecurity threat.<n>We develop PhishOracle, a tool that generates adversarial phishing webpages by embedding diverse phishing features into legitimate webpages.<n>Our findings highlight the vulnerability of phishing detection models to adversarial attacks, emphasizing the need for more robust detection approaches.
  arXiv  Detail & Related papers  (2024-07-29T18:21:34Z)
• BadCLIP: Dual-Embedding Guided Backdoor Attack on Multimodal Contrastive Learning [85.2564206440109]
  This paper reveals the threats in this practical scenario that backdoor attacks can remain effective even after defenses. We introduce the emphtoolns attack, which is resistant to backdoor detection and model fine-tuning defenses.
  arXiv  Detail & Related papers  (2023-11-20T02:21:49Z)
• Defending Pre-trained Language Models as Few-shot Learners against Backdoor Attacks [72.03945355787776]
  We advocate MDP, a lightweight, pluggable, and effective defense for PLMs as few-shot learners. We show analytically that MDP creates an interesting dilemma for the attacker to choose between attack effectiveness and detection evasiveness.
  arXiv  Detail & Related papers  (2023-09-23T04:41:55Z)
• Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection [64.67495502772866]
  Large Language Models (LLMs) are increasingly being integrated into various applications. We show how attackers can override original instructions and employed controls using Prompt Injection attacks. We derive a comprehensive taxonomy from a computer security perspective to systematically investigate impacts and vulnerabilities.
  arXiv  Detail & Related papers  (2023-02-23T17:14:38Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.