arXiv subjects

Fengyuan Xu

Publications and source records attributed to Fengyuan Xu.

At least 19 recordsLinked to original sources

arXiv2026

Usability as a Weapon: Attacking the Safety of LLM-Based Code Generation via Usability Requirements

Large Language Models (LLMs) are increasingly used for automated software development, making their ability to preserve secure coding practices critical. In practice, however, many security requirements are implicit or underspecified, whereas usability requirements are explicit and high-signal. This asymmetry motivates our investigation of usability pressure as a practical attack surface: realistic usability-oriented requirements (e.g., new features, performance constraints, or simplicity demands) can cause coding LLMs to satisfy explicit usability goals while silently dropping implicit security constraints -- a form of reward hacking. We formalize this threat as UPAttack and propose U-SPLOIT, an automated framework to craft UPAttack that (i) selects tasks where a model is initially secure, (ii) synthesizes usability pressures by identifying usability rewards of insecure alternatives across three vectors (Functionality, Implementation, Trade-off), and (iii) verifies security regression via both existing test cases and dynamically generated exploit payloads. Across 75 seed scenarios (25 CWEs x 3 cases), spanning multiple languages (Python, C, and JavaScript), U-SPLOIT achieves attack success rates up to 98.1% on multiple state-of-the-art models (e.g., GPT-5.2-chat and Gemini-3-Flash-Preview).

cs.CR↗

arXiv2026

GTR-Mamba: Geometry-to-Tangent Routing Mamba for Hyperbolic POI Recommendation

Next Point-of-Interest (POI) recommendation is a critical task in modern Location-Based Social Networks (LBSNs), aiming to model the complex decision-making process of human mobility to provide personalized recommendations for a user's next check-in location. Existing hyperbolic POI recommendation models, predominantly based on rotations and graph representations, have been extensively investigated. Although hyperbolic geometry has proven superior in representing hierarchical data with low distortion, current hyperbolic sequence models typically rely on performing recurrence via expensive Möbius operations directly on the manifold. This incurs prohibitive computational costs and numerical instability, rendering them ill-suited for trajectory modeling. To resolve this conflict between geometric representational power and sequential efficiency, we propose GTR-Mamba, a novel framework featuring Geometry-to-Tangent Routing. GTR-Mamba strategically routes complex state transitions to the computationally efficient Euclidean tangent space. Crucially, instead of a static approximation, we introduce a Parallel Transport (PT) mechanism that dynamically aligns tangent spaces along the trajectory. This ensures geometric consistency across recursive updates, effectively bridging the gap between the curved manifold and linear tangent operations. This process is orchestrated by an exogenous spatio-temporal channel, which explicitly modulates the SSM discretization parameters. Extensive experiments on three real-world datasets demonstrate that GTR-Mamba consistently outperforms state-of-the-art baselines in next POI recommendation.

cs.AI↗

arXiv2026

From Transactions to Exploits: Automated PoC Synthesis for Real-World DeFi Attacks

Blockchain systems are increasingly targeted by on-chain attacks that exploit contract vulnerabilities to extract value rapidly and stealthily, making systematic analysis and reproduction highly challenging. In practice, reproducing such attacks requires manually crafting proofs-of-concept (PoCs), a labor-intensive process that demands substantial expertise and scales poorly. In this work, we present the first automated framework for synthesizing verifiable PoCs directly from on-chain attack executions. Our key insight is that attacker logic can be recovered from low-level transaction traces via trace-driven reverse engineering, and then translated into executable exploits by leveraging the code-generation capabilities of large language models (LLMs). To this end, we propose TracExp, which localizes attack-relevant execution contexts from noisy, multi-contract traces and introduces a novel dual-decompiler to transform concrete executions into semantically enriched exploit pseudocode. Guided by this representation, TracExp synthesizes PoCs and refines them to preserve exploitability-relevant semantics. We evaluate TracExp on 321 real-world attacks over the past 20 months. TracExp successfully synthesizes PoCs for 93% of incidents, with 58.78% being directly verifiable, at an average cost of only \$0.07 per case. Moreover, TracExp enabled the release of a large number of previously unavailable PoCs to the community, earning a $900 bounty and demonstrating strong practical impact.

cs.CR↗

arXiv2025

A Systematic Study of Code Obfuscation Against LLM-based Vulnerability Detection

As large language models (LLMs) are increasingly adopted for code vulnerability detection, their reliability and robustness across diverse vulnerability types have become a pressing concern. In traditional adversarial settings, code obfuscation has long been used as a general strategy to bypass auditing tools, preserving exploitability without tampering with the tools themselves. Numerous efforts have explored obfuscation methods and tools, yet their capabilities differ in terms of supported techniques, granularity, and programming languages, making it difficult to systematically assess their impact on LLM-based vulnerability detection. To address this gap, we provide a structured systematization of obfuscation techniques and evaluate them under a unified framework. Specifically, we categorize existing obfuscation methods into three major classes (layout, data flow, and control flow) covering 11 subcategories and 19 concrete techniques. We implement these techniques across four programming languages (Solidity, C, C++, and Python) using a consistent LLM-driven approach, and evaluate their effects on 15 LLMs spanning four model families (DeepSeek, OpenAI, Qwen, and LLaMA), as well as on two coding agents (GitHub Copilot and Codex). Our findings reveal both positive and negative impacts of code obfuscation on LLM-based vulnerability detection, highlighting conditions under which obfuscation leads to performance improvements or degradations. We further analyze these outcomes with respect to vulnerability characteristics, code properties, and model attributes. Finally, we outline several open problems and propose future directions to enhance the robustness of LLMs for real-world vulnerability detection.

cs.CR↗

arXiv2025

Enhancing All-to-X Backdoor Attacks with Optimized Target Class Mapping

Backdoor attacks pose severe threats to machine learning systems, prompting extensive research in this area. However, most existing work focuses on single-target All-to-One (A2O) attacks, overlooking the more complex All-to-X (A2X) attacks with multiple target classes, which are often assumed to have low attack success rates. In this paper, we first demonstrate that A2X attacks are robust against state-of-the-art defenses. We then propose a novel attack strategy that enhances the success rate of A2X attacks while maintaining robustness by optimizing grouping and target class assignment mechanisms. Our method improves the attack success rate by up to 28%, with average improvements of 6.7%, 16.4%, 14.1% on CIFAR10, CIFAR100, and Tiny-ImageNet, respectively. We anticipate that this study will raise awareness of A2X attacks and stimulate further research in this under-explored area. Our code is available at https://github.com/kazefjj/A2X-backdoor .

cs.CR↗

arXiv2025

Revealing Adversarial Smart Contracts through Semantic Interpretation and Uncertainty Estimation

Adversarial smart contracts, mostly on EVM-compatible chains like Ethereum and BSC, are deployed as EVM bytecode to exploit vulnerable smart contracts for financial gain. Detecting such malicious contracts at the time of deployment is an important proactive strategy to prevent losses from victim contracts. It offers a better cost-benefit ratio than detecting vulnerabilities on diverse potential victims. However, existing works are not generic with limited detection types and effectiveness due to imbalanced samples, while the emerging LLM technologies, which show their potential in generalization, have two key problems impeding its application in this task: hard digestion of compiled-code inputs, especially those with task-specific logic, and hard assessment of LLM's certainty in its binary (yes-or-no) answers. Therefore, we propose a generic adversarial smart contracts detection framework FinDet, which leverages LLM with two enhancements addressing the above two problems. FinDet takes as input only the EVM bytecode contracts and identifies adversarial ones among them with high balanced accuracy. The first enhancement extracts concise semantic intentions and high-level behavioral logic from the low-level bytecode inputs, unleashing the LLM reasoning capability restricted by the task input. The second enhancement probes and measures the LLM uncertainty to its multi-round answering to the same query, improving the LLM answering robustness for binary classifications required by the task output. Our comprehensive evaluation shows that FinDet achieves a BAC of 0.9374 and a TPR of 0.9231, significantly outperforming existing baselines. It remains robust under challenging conditions including unseen attack patterns, low-data settings, and feature obfuscation. FinDet detects all 5 public and 20+ unreported adversarial contracts in a 10-day real-world test, confirmed manually.

cs.CR↗

arXiv2025

ConneX: Automatically Resolving Transaction Opacity of Cross-Chain Bridges for Security Analysis

As the Web3 ecosystem evolves toward a multi-chain architecture, cross-chain bridges have become critical infrastructure for enabling interoperability between diverse blockchain networks. However, while connecting isolated blockchains, the lack of cross-chain transaction pairing records introduces significant challenges for security analysis like cross-chain fund tracing, advanced vulnerability detection, and transaction graph-based analysis. To address this gap, we introduce ConneX, an automated and general-purpose system designed to accurately identify corresponding transaction pairs across both ends of cross-chain bridges. Our system leverages Large Language Models (LLMs) to efficiently prune the semantic search space by identifying semantically plausible key information candidates within complex transaction records. Further, it deploys a novel examiner module that refines these candidates by validating them against transaction values, effectively addressing semantic ambiguities and identifying the correct semantics. Extensive evaluations on a dataset of about 500,000 transactions from five major bridge platforms demonstrate that ConneX achieves an average F1 score of 0.9746, surpassing baselines by at least 20.05\%, with good efficiency that reduces the semantic search space by several orders of magnitude (1e10 to less than 100). Moreover, its successful application in tracing illicit funds (including a cross-chain transfer worth $1 million) in real-world hacking incidents underscores its practical utility for enhancing cross-chain security and transparency.

cs.CR↗

arXiv2025

LLM-based Human-like Traffic Simulation for Self-driving Tests

Ensuring realistic traffic dynamics is a prerequisite for simulation platforms to evaluate the reliability of self-driving systems before deployment in the real world. Because most road users are human drivers, reproducing their diverse behaviors within simulators is vital. Existing solutions, however, typically rely on either handcrafted heuristics or narrow data-driven models, which capture only fragments of real driving behaviors and offer limited driving style diversity and interpretability. To address this gap, we introduce HDSim, an HD traffic generation framework that combines cognitive theory with large language model (LLM) assistance to produce scalable and realistic traffic scenarios within simulation platforms. The framework advances the state of the art in two ways: (i) it introduces a hierarchical driver model that represents diverse driving style traits, and (ii) it develops a Perception-Mediated Behavior Influence strategy, where LLMs guide perception to indirectly shape driver actions. Experiments reveal that embedding HDSim into simulation improves detection of safety-critical failures in self-driving systems by up to 68% and yields realism-consistent accident interpretability.

cs.RO↗

arXiv2025

When LLMs Copy to Think: Uncovering Copy-Guided Attacks in Reasoning LLMs

Large Language Models (LLMs) have become integral to automated code analysis, enabling tasks such as vulnerability detection and code comprehension. However, their integration introduces novel attack surfaces. In this paper, we identify and investigate a new class of prompt-based attacks, termed Copy-Guided Attacks (CGA), which exploit the inherent copying tendencies of reasoning-capable LLMs. By injecting carefully crafted triggers into external code snippets, adversaries can induce the model to replicate malicious content during inference. This behavior enables two classes of vulnerabilities: inference length manipulation, where the model generates abnormally short or excessively long reasoning traces; and inference result manipulation, where the model produces misleading or incorrect conclusions. We formalize CGA as an optimization problem and propose a gradient-based approach to synthesize effective triggers. Empirical evaluation on state-of-the-art reasoning LLMs shows that CGA reliably induces infinite loops, premature termination, false refusals, and semantic distortions in code analysis tasks. While highly effective in targeted settings, we observe challenges in generalizing CGA across diverse prompts due to computational constraints, posing an open question for future research. Our findings expose a critical yet underexplored vulnerability in LLM-powered development pipelines and call for urgent advances in prompt-level defense mechanisms.

cs.CR↗

arXiv2025

If LLMs Would Just Look: Simple Line-by-line Checking Improves Vulnerability Localization

The rapid expansion of software systems and the growing number of reported vulnerabilities have emphasized the importance of accurately identifying vulnerable code segments. Traditional methods for vulnerability localization, such as manual code audits or rule-based tools, are often time-consuming and limited in scope, typically focusing on specific programming languages or types of vulnerabilities. In recent years, the introduction of large language models (LLMs) such as GPT and LLaMA has opened new possibilities for automating vulnerability detection. However, while LLMs show promise in this area, they face challenges, particularly in maintaining accuracy over longer code contexts. This paper introduces LOVA, a novel framework leveraging the self-attention mechanisms inherent in LLMs to enhance vulnerability localization. Our key insight is that self-attention mechanisms assign varying importance to different parts of the input, making it possible to track how much attention the model focuses on specific lines of code. In the context of vulnerability localization, the hypothesis is that vulnerable lines of code will naturally attract higher attention weights because they have a greater influence on the model's output. By systematically tracking changes in attention weights and focusing on specific lines of code, LOVA improves the precision of identifying vulnerable lines across various programming languages. Through rigorous experimentation and evaluation, we demonstrate that LOVA significantly outperforms existing LLM-based approaches, achieving up to a 5.3x improvement in F1-scores. LOVA also demonstrated strong scalability, with up to a 14.6x improvement in smart contract vulnerability localization across languages like C, Python, Java, and Solidity. Its robustness was proven through consistent performance across different LLM architectures.

cs.CR↗

arXiv2025

Everything You Wanted to Know About LLM-based Vulnerability Detection But Were Afraid to Ask

Large Language Models are a promising tool for automated vulnerability detection, thanks to their success in code generation and repair. However, despite widespread adoption, a critical question remains: Are LLMs truly effective at detecting real-world vulnerabilities? Current evaluations, which often assess models on isolated functions or files, ignore the broader execution and data-flow context essential for understanding vulnerabilities. This oversight leads to two types of misleading outcomes: incorrect conclusions and flawed rationales, collectively undermining the reliability of prior assessments. Therefore, in this paper, we challenge three widely held community beliefs: that LLMs are (i) unreliable, (ii) insensitive to code patches, and (iii) performance-plateaued across model scales. We argue that these beliefs are artifacts of context-deprived evaluations. To address this, we propose CORRECT (Context-Rich Reasoning Evaluation of Code with Trust), a new evaluation framework that systematically incorporates contextual information into LLM-based vulnerability detection. We construct a context-rich dataset of 2,000 vulnerable-patched program pairs spanning 99 CWEs and evaluate 13 LLMs across four model families. Our framework elicits both binary predictions and natural-language rationales, which are further validated using LLM-as-a-judge techniques. Our findings overturn existing misconceptions. When provided with sufficient context, SOTA LLMs achieve significantly improved performance (e.g., 0.7 F1-score on key CWEs), with 0.8 precision. We show that most false positives stem from reasoning errors rather than misclassification, and that while model and test-time scaling improve performance, they introduce diminishing returns and trade-offs in recall. Finally, we uncover new flaws in current LLM-based detection systems, such as limited generalization and overthinking biases.

cs.CR↗

arXiv2025

Unleashing the Power of LLM to Infer State Machine from the Protocol Implementation

State machines are essential for enhancing protocol analysis to identify vulnerabilities. However, inferring state machines from network protocol implementations is challenging due to complex code syntax and semantics. Traditional dynamic analysis methods often miss critical state transitions due to limited coverage, while static analysis faces path explosion issues. To overcome these challenges, we introduce a novel state machine inference approach utilizing Large Language Models (LLMs), named ProtocolGPT. This method employs retrieval augmented generation technology to enhance a pre-trained model with specific knowledge from protocol implementations. Through effective prompt engineering, we accurately identify and infer state machines. To the best of our knowledge, our approach represents the first state machine inference that leverages the source code of protocol implementations. Our evaluation of six protocol implementations shows that our method achieves a precision of over 90%, outperforming the baselines by more than 30%. Furthermore, integrating our approach with protocol fuzzing improves coverage by more than 20% and uncovers two 0-day vulnerabilities compared to baseline methods.

cs.CR↗

arXiv2025

Making Every Frame Matter: Continuous Activity Recognition in Streaming Video via Adaptive Video Context Modeling

Video activity recognition has become increasingly important in robots and embodied AI. Recognizing continuous video activities poses considerable challenges due to the fast expansion of streaming video, which contains multi-scale and untrimmed activities. We introduce a novel system, CARS, to overcome these issues through adaptive video context modeling. Adaptive video context modeling refers to selectively maintaining activity-related features in temporal and spatial dimensions. CARS has two key designs. The first is an activity spatial feature extraction by eliminating irrelevant visual features while maintaining recognition accuracy. The second is an activity-aware state update introducing dynamic adaptability to better preserve the video context for multi-scale activity recognition. Our CARS runs at speeds $>$30 FPS on typical edge devices and outperforms all baselines by 1.2\% to 79.7\% in accuracy. Moreover, we explore applying CARS to a large video model as a video encoder. Experimental results show that our CARS can result in a 0.46-point enhancement (on a 5-point scale) on the in-distribution video activity dataset, and an improvement ranging from 1.19\% to 4\% on zero-shot video activity datasets.

cs.CV↗

arXiv2024

Privacy-Preserving Federated Learning via Dataset Distillation

Federated Learning (FL) allows users to share knowledge instead of raw data to train a model with high accuracy. Unfortunately, during the training, users lose control over the knowledge shared, which causes serious data privacy issues. We hold that users are only willing and need to share the essential knowledge to the training task to obtain the FL model with high accuracy. However, existing efforts cannot help users minimize the shared knowledge according to the user intention in the FL training procedure. This work proposes FLiP, which aims to bring the principle of least privilege (PoLP) to FL training. The key design of FLiP is applying elaborate information reduction on the training data through a local-global dataset distillation design. We measure the privacy performance through attribute inference and membership inference attacks. Extensive experiments show that FLiP strikes a good balance between model accuracy and privacy protection.

cs.LG↗

arXiv2024

Training Data Attribution: Was Your Model Secretly Trained On Data Created By Mine?

The emergence of text-to-image models has recently sparked significant interest, but the attendant is a looming shadow of potential infringement by violating the user terms. Specifically, an adversary may exploit data created by a commercial model to train their own without proper authorization. To address such risk, it is crucial to investigate the attribution of a suspicious model's training data by determining whether its training data originates, wholly or partially, from a specific source model. To trace the generated data, existing methods require applying extra watermarks during either the training or inference phases of the source model. However, these methods are impractical for pre-trained models that have been released, especially when model owners lack security expertise. To tackle this challenge, we propose an injection-free training data attribution method for text-to-image models. It can identify whether a suspicious model's training data stems from a source model, without additional modifications on the source model. The crux of our method lies in the inherent memorization characteristic of text-to-image models. Our core insight is that the memorization of the training dataset is passed down through the data generated by the source model to the model trained on that data, making the source model and the infringing model exhibit consistent behaviors on specific samples. Therefore, our approach involves developing algorithms to uncover these distinct samples and using them as inherent watermarks to verify if a suspicious model originates from the source model. Our experiments demonstrate that our method achieves an accuracy of over 80\% in identifying the source of a suspicious model's training data, without interfering the original training or generation process of the source model.

cs.CV↗

arXiv2024

CoAst: Validation-Free Contribution Assessment for Federated Learning based on Cross-Round Valuation

In the federated learning (FL) process, since the data held by each participant is different, it is necessary to figure out which participant has a higher contribution to the model performance. Effective contribution assessment can help motivate data owners to participate in the FL training. Research works in this field can be divided into two directions based on whether a validation dataset is required. Validation-based methods need to use representative validation data to measure the model accuracy, which is difficult to obtain in practical FL scenarios. Existing validation-free methods assess the contribution based on the parameters and gradients of local models and the global model in a single training round, which is easily compromised by the stochasticity of model training. In this work, we propose CoAst, a practical method to assess the FL participants' contribution without access to any validation data. The core idea of CoAst involves two aspects: one is to only count the most important part of model parameters through a weights quantization, and the other is a cross-round valuation based on the similarity between the current local parameters and the global parameter updates in several subsequent communication rounds. Extensive experiments show that CoAst has comparable assessment reliability to existing validation-based methods and outperforms existing validation-free methods.

cs.LG↗

arXiv2023

Manipulating Transfer Learning for Property Inference

Transfer learning is a popular method for tuning pretrained (upstream) models for different downstream tasks using limited data and computational resources. We study how an adversary with control over an upstream model used in transfer learning can conduct property inference attacks on a victim's tuned downstream model. For example, to infer the presence of images of a specific individual in the downstream training set. We demonstrate attacks in which an adversary can manipulate the upstream model to conduct highly effective and specific property inference attacks (AUC score $> 0.9$), without incurring significant performance loss on the main task. The main idea of the manipulation is to make the upstream model generate activations (intermediate features) with different distributions for samples with and without a target property, thus enabling the adversary to distinguish easily between downstream models trained with and without training examples that have the target property. Our code is available at https://github.com/yulongt23/Transfer-Inference.

cs.LG↗

arXiv2022

LEAP: TrustZone Based Developer-Friendly TEE for Intelligent Mobile Apps

ARM TrustZone is widely deployed on commercial-off-the-shelf mobile devices for secure execution. However, many Apps cannot enjoy this feature because it brings many constraints to App developers. Previous works have been proposed to build a secure execution environment for developers on top of TrustZone. Unfortunately, these works are still not fully-fledged solutions for mobile Apps, especially for emerging intelligent Apps. To this end, we propose LEAP, which is a lightweight developer-friendly TEE solution for mobile Apps. LEAP enables isolated codes to execute in parallel and access peripheral (e.g., mobile GPUs) with ease, flexibly manages system resources upon different workloads, and offers the auto DevOps tool to help developers prepare the codes running on it. We implement the LEAP prototype on the off-the-shelf ARM platform and conduct extensive experiments on it. The experimental results show that Apps can be adapted to run with LEAP easily and efficiently. Compared to the state-of-the-art work along this research line, LEAP can achieve an average 3.57x speedup in supporting intelligent Apps using mobile GPU acceleration.

cs.CR↗