arXiv subjects

Haojin Zhu

Publications and source records attributed to Haojin Zhu.

At least 19 recordsLinked to original sources

arXiv2026

PiMRef: Deducing Ever-evolving Spear-phishing Emails with Knowledge Base Invariants

Phishing email is a critical step in the cybercrime kill chain due to the high reachability of victims' email accounts and the low cost of launching phishing campaigns. This ever-evolving nature of phishing emails makes traditional rule-based and feature-engineering-based phishing email detectors fight an uphill battle in the cat-and-mouse game of defense and attack. In this work, we show that, large language models (LLMs) can be effectively exploited to generate profile-grounded spear-phishing, compromising major paradigms of phishing email detectors. To defend against such LLM-based spear-phishing attacks, we propose PiMRef, the first reference-based solution to detect ever-evolving phishing emails using knowledge-based invariants, targeting the identity-impersonation attacks that characterize spear-phishing. Our rationale lies in the fact that convincing phishing emails often include ``disprovable claims'', which contradict certain real-world facts. Technically, given an email, PiMRef (i) discovers the claimed identity of the sender, (ii) verifies the sender's email domain against a dynamically expandable knowledge base, and (iii) infers call-to-action instructions that encourage next-step engagement. Compared to existing baselines such as D-Fence, HelpHed, and ChatSpamDetector, PimRef reduces the false-positive rate to 0.81\% while maintaining a recall of 90.7\%-93.1\% on conventional phishing benchmarks such as Nazario and PhishPot. On SpearMail, our newly constructed benchmark of 14,672 LLM-generated spear-phishing emails targeting 681 public profiles, PimRef reaches a recall of 86.4\% without incurring additional false positives. Our code is publicly available at https://github.com/code-philia/PhishEmail.

cs.CR↗

arXiv2026

Security of World-Model-Based Embodied AI: A Lifecycle of Threats, Defenses, and Evaluation

World models give embodied AI a predictive core: they compress observations into states, simulate action-conditioned futures, and enable planning beyond reactive control. This predictive layer, however, opens a new security boundary-compromise can propagate from data, sensors, prompts, or feedback into physical action. Rather than treating world models as an isolated component, this survey traces threats across their entire lifecycle-from data construction and representation learning, through state grounding and imagination, to trajectory evaluation, execution, and long-term adaptation via memory and tools. We show that familiar attack families: poisoning, backdoors, adversarial examples, sensor spoofing, prompt injection, trajectory manipulation, and supply-chain attacks take on distinct meanings when they corrupt world states, learned dynamics, affordance estimates, or safety costs. We also highlight a duality: world models can serve as runtime safety shields, yet when compromised or over-trusted they generate predictive safety illusions. The survey offers a lifecycle taxonomy, maps existing attacks to world-model security properties, outlines evaluation protocols for safety failures, and structures defenses across provenance, robust grounding, uncertainty-aware prediction, trajectory gating, feedback auditing, and deployment assurance.

cs.CR↗

arXiv2026

KQFuzz: Knowledge-Guided Fuzzing for Quantum Libraries via Large Language Models

As quantum computing continually improves, ensuring the reliability and correctness of quantum libraries has become increasingly critical. To this end, many LLM-based fuzzing approaches towards quantum libraries have been proposed to uncover potential bugs. However, these methods still suffer from limitations such as insufficient flexibility and low efficiency, which hinder the progress of the quantum computing field. To address these challenges, we propose KQFuzz, a novel knowledge-guided fuzzer for quantum libraries. It leverages comprehensive codebase knowledge to ground LLM-based test generation, synergizing this with fitness-guided evaluation and two-level mutations to explore complex execution paths and trigger potential bugs. Firstly, KQFuzz introduces a novel prompting scheme tailored to quantum programs, which strategically incorporates knowledge of the codebase to efficiently generate high-quality quantum seed programs. Moreover, we develop evaluation and mutation strategies to handle the generated seed programs, facilitating efficient fuzzing execution while further enriching the diversity of the resulting test cases. We implement KQFuzz and conduct fuzzing on three popular quantum libraries, including Qiskit, PennyLane, and Cirq. Experimental results demonstrate that our approach significantly outperforms other state-of-the-art methods, with coverage improved by up to 18.44%. During the development of KQFuzz, we discovered 13 bugs, all of which have been confirmed and 12 have already been fixed by the developers.

cs.SE↗

arXiv2026

SlowBA: An efficiency backdoor attack towards VLM-based GUI agents

Modern vision-language-model (VLM) based graphical user interface (GUI) agents are expected not only to execute actions accurately but also to respond to user instructions with low latency. While existing research on GUI-agent security mainly focuses on manipulating action correctness, the security risks related to response efficiency remain largely unexplored. In this paper, we introduce SlowBA, a novel backdoor attack that targets the responsiveness of VLM-based GUI agents. The key idea is to manipulate response latency by inducing excessively long reasoning chains under specific trigger patterns. To achieve this, we propose a two-stage reward-level backdoor injection (RBI) strategy that first aligns the long-response format and then learns trigger-aware activation through reinforcement learning. In addition, we design realistic pop-up windows as triggers that naturally appear in GUI environments, improving the stealthiness of the attack. Extensive experiments across multiple datasets and baselines demonstrate that SlowBA can significantly increase response length and latency while largely preserving task accuracy. The attack remains effective even with a small poisoning ratio and under several defense settings. These findings reveal a previously overlooked security vulnerability in GUI agents and highlight the need for defenses that consider both action correctness and response efficiency. Code can be found in https://github.com/tu-tuing/SlowBA.

cs.CR↗

arXiv2026

DIPBox: A Multi-scale Testing Framework for Tracking Dataset Regeneration

Training datasets have tremendous proprietary value and are vulnerable to unauthorized copying. Existing defenses mainly focus on tracking individual data points, but pay little attention to the threat of dataset regeneration. Through a measurement study of public tumor datasets, we identify substantial real-world partial-dataset replication, raising concerns about potential license noncompliance. To counter the challenge of tracking previously unknown adversarial regeneration, our key insight is that regeneration that preserves model utility inevitably preserves measurable signals across multiple feature scales. We categorize these dataset features into sample-, set-, and distribution-level features and design four similarity metrics to accurately identify regeneration. Based on these metrics, we develop DIPBox, which to our knowledge is the first testing framework that tracks regeneration suspects via multi-scale similarity testing across a spectrum of defender access settings, from limited to full information. We further provide a learning-theoretic analysis that justifies these multi-scale metrics and formalizes an inherent utility--divergence trade-off, implying fundamental limits on evasive regeneration. Extensive experiments on 16 vision and text base datasets, 320 regenerated datasets, and 590 derived models validate that DIPBox outperforms previous solutions while characterizing its robustness and limits under three adaptive attacks.

cs.CR↗

arXiv2026

Patronus: Identifying and Mitigating Transferable Backdoors in Pre-trained Language Models

The ``Pre-train, then fine-tune'' paradigm has revolutionized Natural Language Processing (NLP). In this context, transferable backdoors pose a severe threat to the Pre-trained Language Models (PLMs) supply chain, yet defensive research remains nascent, primarily relying on detecting anomalies in the output feature space. We identify a critical flaw that fine-tuning on downstream tasks inevitably modifies model parameters, shifting the output distribution and rendering pre-computed defense ineffective. To address this, we propose Patronus, a novel defense framework that shifts the defensive focus from output features to input-side invariance, exploiting the fact that adversarial triggers remain constant even as model weights change. To overcome the convergence challenges of discrete text optimization, Patronus introduces a multi-trigger contrastive search algorithm that effectively bridges gradient-based optimization with contrastive learning objectives. Furthermore, we employ a dual-stage mitigation strategy combining real-time input monitoring with model purification via adversarial training. Extensive experiments across 15 PLMs and nine tasks demonstrate that Patronus achieves $\geq98.3\%$ backdoor detection recall and reduces attack success rates to clean settings, significantly outperforming all state-of-the-art baselines in all settings. Code is available at https://github.com/zth855/Patronus.

cs.CR↗

arXiv2026

Trojan's Whisper: Stealthy Manipulation of OpenClaw through Injected Bootstrapped Guidance

Autonomous coding agents are increasingly integrated into software development workflows, offering capabilities that extend beyond code suggestion to active system interaction and environment management. OpenClaw, a representative platform in this emerging paradigm, introduces an extensible skill ecosystem that allows third-party developers to inject behavioral guidance through lifecycle hooks during agent initialization. While this design enhances automation and customization, it also opens a novel and unexplored attack surface. In this paper, we identify and systematically characterize guidance injection, a stealthy attack vector that embeds adversarial operational narratives into bootstrap guidance files. Unlike traditional prompt injection, which relies on explicit malicious instructions, guidance injection manipulates the agent's reasoning context by framing harmful actions as routine best practices. These narratives are automatically incorporated into the agent's interpretive framework and influence future task execution without raising suspicion.We construct 26 malicious skills spanning 13 attack categories including credential exfiltration, workspace destruction, privilege escalation, and persistent backdoor installation. We evaluate them using ORE-Bench, a realistic developer workspace benchmark we developed. Across 52 natural user prompts and six state-of-the-art LLM backends, our attacks achieve success rates from 16.0% to 64.2%, with the majority of malicious actions executed autonomously without user confirmation. Furthermore, 94% of our malicious skills evade detection by existing static and LLM-based scanners. Our findings reveal fundamental tensions in the design of autonomous agent ecosystems and underscore the urgent need for defenses based on capability isolation, runtime policy enforcement, and transparent guidance provenance.

cs.CR↗

arXiv2026

EmbTracker: Traceable Black-box Watermarking for Federated Language Models

Federated Language Model (FedLM) allows a collaborative learning without sharing raw data, yet it introduces a critical vulnerability, as every untrustworthy client may leak the received functional model instance. Current watermarking schemes for FedLM often require white-box access and client-side cooperation, providing only group-level proof of ownership rather than individual traceability. We propose EmbTracker, a server-side, traceable black-box watermarking framework specifically designed for FedLMs. EmbTracker achieves black-box verifiability by embedding a backdoor-based watermark detectable through simple API queries. Client-level traceability is realized by injecting unique identity-specific watermarks into the model distributed to each client. In this way, a leaked model can be attributed to a specific culprit, ensuring robustness even against non-cooperative participants. Extensive experiments on various language and vision-language models demonstrate that EmbTracker achieves robust traceability with verification rates near 100\%, high resilience against removal attacks (fine-tuning, pruning, quantization), and negligible impact on primary task performance (typically within 1-2\%).

cs.CR↗

arXiv2025

Your Microphone Array Retains Your Identity: A Robust Voice Liveness Detection System for Smart Speakers

Though playing an essential role in smart home systems, smart speakers are vulnerable to voice spoofing attacks. Passive liveness detection, which utilizes only the collected audio rather than the deployed sensors to distinguish between live-human and replayed voices, has drawn increasing attention. However, it faces the challenge of performance degradation under the different environmental factors as well as the strict requirement of the fixed user gestures. In this study, we propose a novel liveness feature, array fingerprint, which utilizes the microphone array inherently adopted by the smart speaker to determine the identity of collected audios. Our theoretical analysis demonstrates that by leveraging the circular layout of microphones, compared with existing schemes, array fingerprint achieves a more robust performance under the environmental change and user's movement. Then, to leverage such a fingerprint, we propose ARRAYID, a lightweight passive detection scheme, and elaborate a series of features working together with array fingerprint. Our evaluation on the dataset containing 32,780 audio samples and 14 spoofing devices shows that ARRAYID achieves an accuracy of 99.84%, which is superior to existing passive liveness detection schemes.

cs.CR↗

arXiv2025

Leveraging Scene Context with Dual Networks for Sequential User Behavior Modeling

Modeling sequential user behaviors for future behavior prediction is crucial in improving user's information retrieval experience. Recent studies highlight the importance of incorporating contextual information to enhance prediction performance. One crucial but usually neglected contextual information is the scene feature which we define as sub-interfaces within an app, created by developers to provide specific functionalities, such as ``text2product search" and ``live" modules in e-commence apps. Different scenes exhibit distinct functionalities and usage habits, leading to significant distribution gap in user engagement across them. Popular sequential behavior models either ignore the scene feature or merely use it as attribute embeddings, which cannot effectively capture the dynamic interests and interplay between scenes and items when modeling user sequences. In this work, we propose a novel Dual Sequence Prediction networks (DSPnet) to effectively capture the dynamic interests and interplay between scenes and items for future behavior prediction. DSPnet consists of two parallel networks dedicated to learn users' dynamic interests over items and scenes, and a sequence feature enhancement module to capture the interplay for enhanced future behavior prediction. Further, we introduce a Conditional Contrastive Regularization (CCR) loss to capture the invariance of similar historical sequences. Theoretical analysis suggests that DSPnet is a principled way to learn the joint relationships between scene and item sequences. Extensive experiments are conducted on one public benchmark and two collected industrial datasets. The method has been deployed online in our system, bringing a 0.04 point increase in CTR, 0.78\% growth in deals, and 0.64\% rise in GMV. The codes are available at this anonymous github: \textcolor{blue}{https://anonymous.4open.science/r/DSPNet-ForPublish-2506/}.

cs.IR↗

arXiv2025

Step-wise Adaptive Integration of Supervised Fine-tuning and Reinforcement Learning for Task-Specific LLMs

Large language models (LLMs) excel at mathematical reasoning and logical problem-solving. The current popular training paradigms primarily use supervised fine-tuning (SFT) and reinforcement learning (RL) to enhance the models' reasoning abilities. However, when using SFT or RL alone, there are respective challenges: SFT may suffer from overfitting, while RL is prone to mode collapse. The state-of-the-art methods have proposed hybrid training schemes. However, static switching faces challenges such as poor generalization across different tasks and high dependence on data quality. In response to these challenges, inspired by the curriculum learning-quiz mechanism in human reasoning cultivation, We propose SASR, a step-wise adaptive hybrid training framework that theoretically unifies SFT and RL and dynamically balances the two throughout optimization. SASR uses SFT for initial warm-up to establish basic reasoning skills, and then uses an adaptive dynamic adjustment algorithm based on gradient norm and divergence relative to the original distribution to seamlessly integrate SFT with the online RL method GRPO. By monitoring the training status of LLMs and adjusting the training process in sequence, SASR ensures a smooth transition between training schemes, maintaining core reasoning abilities while exploring different paths. Experimental results demonstrate that SASR outperforms SFT, RL, and static hybrid training methods.

cs.LG↗

arXiv2025

Depth Gives a False Sense of Privacy: LLM Internal States Inversion

Large Language Models (LLMs) are increasingly integrated into daily routines, yet they raise significant privacy and safety concerns. Recent research proposes collaborative inference, which outsources the early-layer inference to ensure data locality, and introduces model safety auditing based on inner neuron patterns. Both techniques expose the LLM's Internal States (ISs), which are traditionally considered irreversible to inputs due to optimization challenges and the highly abstract representations in deep layers. In this work, we challenge this assumption by proposing four inversion attacks that significantly improve the semantic similarity and token matching rate of inverted inputs. Specifically, we first develop two white-box optimization-based attacks tailored for low-depth and high-depth ISs. These attacks avoid local minima convergence, a limitation observed in prior work, through a two-phase inversion process. Then, we extend our optimization attack under more practical black-box weight access by leveraging the transferability between the source and the derived LLMs. Additionally, we introduce a generation-based attack that treats inversion as a translation task, employing an inversion model to reconstruct inputs. Extensive evaluation of short and long prompts from medical consulting and coding assistance datasets and 6 LLMs validates the effectiveness of our inversion attacks. Notably, a 4,112-token long medical consulting prompt can be nearly perfectly inverted with 86.88 F1 token matching from the middle layer of Llama-3 model. Finally, we evaluate four practical defenses that we found cannot perfectly prevent ISs inversion and draw conclusions for future mitigation design.

cs.CR↗

arXiv2025

Model Inversion in Split Learning for Personalized LLMs: New Insights from Information Bottleneck Theory

Personalized Large Language Models (LLMs) have become increasingly prevalent, showcasing the impressive capabilities of models like GPT-4. This trend has also catalyzed extensive research on deploying LLMs on mobile devices. Feasible approaches for such edge-cloud deployment include using split learning. However, previous research has largely overlooked the privacy leakage associated with intermediate representations transmitted from devices to servers. This work is the first to identify model inversion attacks in the split learning framework for LLMs, emphasizing the necessity of secure defense. For the first time, we introduce mutual information entropy to understand the information propagation of Transformer-based LLMs and assess privacy attack performance for LLM blocks. To address the issue of representations being sparser and containing less information than embeddings, we propose a two-stage attack system in which the first part projects representations into the embedding space, and the second part uses a generative model to recover text from these embeddings. This design breaks down the complexity and achieves attack scores of 38%-75% in various scenarios, with an over 60% improvement over the SOTA. This work comprehensively highlights the potential privacy risks during the deployment of personalized LLMs on the edge side.

cs.LG↗

arXiv2024

The Philosopher's Stone: Trojaning Plugins of Large Language Models

Open-source Large Language Models (LLMs) have recently gained popularity because of their comparable performance to proprietary LLMs. To efficiently fulfill domain-specialized tasks, open-source LLMs can be refined, without expensive accelerators, using low-rank adapters. However, it is still unknown whether low-rank adapters can be exploited to control LLMs. To address this gap, we demonstrate that an infected adapter can induce, on specific triggers,an LLM to output content defined by an adversary and to even maliciously use tools. To train a Trojan adapter, we propose two novel attacks, POLISHED and FUSION, that improve over prior approaches. POLISHED uses a superior LLM to align naïvely poisoned data based on our insight that it can better inject poisoning knowledge during training. In contrast, FUSION leverages a novel over-poisoning procedure to transform a benign adapter into a malicious one by magnifying the attention between trigger and target in model weights. In our experiments, we first conduct two case studies to demonstrate that a compromised LLM agent can use malware to control the system (e.g., a LLM-driven robot) or to launch a spear-phishing attack. Then, in terms of targeted misinformation, we show that our attacks provide higher attack effectiveness than the existing baseline and, for the purpose of attracting downloads, preserve or improve the adapter's utility. Finally, we designed and evaluated three potential defenses. However, none proved entirely effective in safeguarding against our attacks, highlighting the need for more robust defenses supporting a secure LLM supply chain.

cs.CR↗

arXiv2024

VPVet: Vetting Privacy Policies of Virtual Reality Apps

Virtual reality (VR) apps can harvest a wider range of user data than web/mobile apps running on personal computers or smartphones. Existing law and privacy regulations emphasize that VR developers should inform users of what data are collected/used/shared (CUS) through privacy policies. However, privacy policies in the VR ecosystem are still in their early stages, and many developers fail to write appropriate privacy policies that comply with regulations and meet user expectations. In this paper, we propose VPVet to automatically vet privacy policy compliance issues for VR apps. VPVet first analyzes the availability and completeness of a VR privacy policy and then refines its analysis based on three key criteria: granularity, minimization, and consistency of CUS statements. Our study establishes the first and currently largest VR privacy policy dataset named VRPP, consisting of privacy policies of 11,923 different VR apps from 10 mainstream platforms. Our vetting results reveal severe privacy issues within the VR ecosystem, including the limited availability and poor quality of privacy policies, along with their coarse granularity, lack of adaptation to VR traits and the inconsistency between CUS statements in privacy policies and their actual behaviors. We open-source VPVet system along with our findings at repository https://github.com/kalamoo/PPAudit, aiming to raise awareness within the VR community and pave the way for further research in this field.

cs.CR↗

arXiv2024

A Duty to Forget, a Right to be Assured? Exposing Vulnerabilities in Machine Unlearning Services

The right to be forgotten requires the removal or "unlearning" of a user's data from machine learning models. However, in the context of Machine Learning as a Service (MLaaS), retraining a model from scratch to fulfill the unlearning request is impractical due to the lack of training data on the service provider's side (the server). Furthermore, approximate unlearning further embraces a complex trade-off between utility (model performance) and privacy (unlearning performance). In this paper, we try to explore the potential threats posed by unlearning services in MLaaS, specifically over-unlearning, where more information is unlearned than expected. We propose two strategies that leverage over-unlearning to measure the impact on the trade-off balancing, under black-box access settings, in which the existing machine unlearning attacks are not applicable. The effectiveness of these strategies is evaluated through extensive experiments on benchmark datasets, across various model architectures and representative unlearning approaches. Results indicate significant potential for both strategies to undermine model efficacy in unlearning scenarios. This study uncovers an underexplored gap between unlearning and contemporary MLaaS, highlighting the need for careful considerations in balancing data unlearning, model utility, and security.

cs.CR↗

arXiv2023

Mate! Are You Really Aware? An Explainability-Guided Testing Framework for Robustness of Malware Detectors

Numerous open-source and commercial malware detectors are available. However, their efficacy is threatened by new adversarial attacks, whereby malware attempts to evade detection, e.g., by performing feature-space manipulation. In this work, we propose an explainability-guided and model-agnostic testing framework for robustness of malware detectors when confronted with adversarial attacks. The framework introduces the concept of Accrued Malicious Magnitude (AMM) to identify which malware features could be manipulated to maximize the likelihood of evading detection. We then use this framework to test several state-of-the-art malware detectors' abilities to detect manipulated malware. We find that (i) commercial antivirus engines are vulnerable to AMM-guided test cases; (ii) the ability of a manipulated malware generated using one detector to evade detection by another detector (i.e., transferability) depends on the overlap of features with large AMM values between the different detectors; and (iii) AMM values effectively measure the fragility of features (i.e., capability of feature-space manipulation to flip the prediction results) and explain the robustness of malware detectors facing evasion attacks. Our findings shed light on the limitations of current malware detectors, as well as how they can be improved.

cs.CR↗

arXiv2022

Fingerprinting Deep Neural Networks Globally via Universal Adversarial Perturbations

In this paper, we propose a novel and practical mechanism which enables the service provider to verify whether a suspect model is stolen from the victim model via model extraction attacks. Our key insight is that the profile of a DNN model's decision boundary can be uniquely characterized by its Universal Adversarial Perturbations (UAPs). UAPs belong to a low-dimensional subspace and piracy models' subspaces are more consistent with victim model's subspace compared with non-piracy model. Based on this, we propose a UAP fingerprinting method for DNN models and train an encoder via contrastive learning that takes fingerprint as inputs, outputs a similarity score. Extensive studies show that our framework can detect model IP breaches with confidence > 99.99 within only 20 fingerprints of the suspect model. It has good generalizability across different model architectures and is robust against post-modifications on stolen models.

cs.CR↗