arXiv subjects

Xusheng Xiao

Publications and source records attributed to Xusheng Xiao.

At least 19 recordsLinked to original sources

arXiv2026

Your Harness is Not Secure: Benchmarking Real-world Threat of Command Line Interface Agent

Command-line interface (CLI) agents powered by large language models (LLMs) can interpret natural-language requests, plan multi-step tasks, execute shell commands, and modify files and system state. As these agents are increasingly used for operating-system (OS) workflows, it is important to evaluate whether they can be misused to carry out security-relevant operations. Existing benchmarks often lack an attacker-knowledge model grounded in tactics, techniques, and procedures (TTPs), provide limited coverage of end-to-end kill chains, rely on simplified single-host environments, or use LLM-as-a-judge for success evaluation. We introduce AdvCLI, an MITRE ATT&CK-aligned benchmark for evaluating OS-level misuse risks of CLI agents in a controlled multi-host sandbox. AdvCLI contains 140 tasks: 40 direct malicious requests, 74 TTP-based tasks, and 26 end-to-end kill chains. Each task is paired with deterministic hard-coded verification protocols that check whether the requested OS-level effect is realized. We evaluate seven CLI agents and products built on nine foundation models, including ReAct, OpenClaw, OpenAI Agent SDK, Claude Code, Gemini CLI, Cursor CLI, and Cursor IDE. Results show that current CLI agents frequently proceed beyond refusal and can complete a non-negligible fraction of malicious OS-level tasks, especially when requests include TTP-style attacker knowledge. AdvCLI provides a reproducible testbed for evaluating these risks and for developing stronger safety mechanisms for tool-using CLI agents in the future.

cs.CR↗

arXiv2026

The Anatomy of Scam Scenarios: Large-Scale Characterization and Conversation-Aware Detection

Online scams have become a pervasive global threat, causing substantial financial, psychological, and operational harm. Scammers embed psychological techniques (PTs) within reusable operational schemes to scale scam campaigns with minimal adaptation. However, existing studies often analyze PTs as isolated features, overlooking the recurring scam scenarios in which they are systematically deployed. To address this gap, we first conduct a large-scale empirical study to jointly characterize scam scenarios and their associated PTs. Specifically, we develop a data-driven pipeline to derive a hierarchical taxonomy of scam scenarios, consisting of 18 fine-grained scenarios grouped into 6 high-level tactics based on their PT profiles. Furthermore, to transfer this scenario-level knowledge to practical defense, we design a conversation-aware scam scenario detection approach for financial-institution customer interactions, enabling timely warning and intervention. Our study on 102,054 real-world scam incident reports, spanning 2024-02-01 to 2025-10-31, reveals that PT usage is significantly associated with scam scenarios. We further show that scammers organize scenarios around different operational goals, such as broad victim exposure, high victim conversion, and high-value extraction, and reuse infrastructure, including IP addresses, domains, email addresses, and phone numbers, to launch coordinated campaigns at scale. Evaluation on

cs.CR↗

arXiv2026

GRID: Graph Representation of Intelligence Data for Security Text Knowledge Graph Construction

Security knowledge graphs can provide computable external memory for security agents, but constructing them from long-form cyber threat intelligence (CTI) remains difficult: LLMs often lack grounded security-domain knowledge, and end-to-end document-to-graph training is hard to supervise with cheap, stable rewards. We present GRID (Graph Representation of Intelligence Data), an end-to-end framework for security text knowledge graph construction. GRID first builds security-domain supervision from CTI articles by creating traceable article-graph alignments through graph extraction and knowledge-graph-conditioned text revision. It then turns document-to-graph learning into a scripted task bank combining four-option multi-select questions with triple-level regex matching targets, yielding more stable task-specific rewards than repeatedly scoring full graph outputs with an LLM judge. Using this supervision pipeline, we train two Qwen3-4B-Instruct-2507-based 4B extractors: a primary Task-bank Reward model and a secondary End2End Reward model with LLM-as-judge precision/recall rewards. On 249 CTI articles from GRID, CASIE, CTINexus, MalKG, and SecureNLP, the Task-bank Reward model with the ontology-guided GRID extraction pipeline reaches 84.62% source-averaged precision, 64.91% source-averaged recall, and 68.53% Avg F1, achieving the best source-averaged recall and near-top Avg F1 with lower token usage and deployment cost. The End2End Reward model reaches 76.91% precision, 53.85% recall, and 58.06% Avg F1. Further analyses show that task-bank rewards can be built once offline and reused across later post-training runs, outperforming online End2End LLM-as-judge reward and weaker alternatives such as Choice-only Reward and End2End SFT without RL.

cs.AI↗

arXiv2026

PreScam: A Benchmark for Predicting Scam Progression from Early Conversations

Conversational scams, such as romance and investment scams, are emerging as a major form of online fraud. Unlike one-shot scam lures such as fake lottery or unpaid toll messages, they unfold through multi-turn conversations in which scammers gradually manipulate victims using evolving psychological techniques. However, existing research mainly focuses on static scam detection or synthetic scams, leaving open whether language models can understand how real-world scams progress over time. We introduce PreScam, a benchmark for modeling scam progression from early conversations. Built from user-submitted scam reports, PreScam filters and structures 177,989 raw reports into 11,573 conversational scam instances spanning 20 scam categories. Each instance is hierarchically structured according to the scam lifecycle defined by the proposed scam kill chain, and further annotated at the turn level with scammer psychological actions and victim responses. We benchmark models on two tasks: real-time termination prediction, which estimates whether a conversation is approaching the termination stage, and scammer action prediction, which forecasts the scammer's subsequent actions. Results show a clear gap between surface-level fluency and progression modeling: supervised encoders substantially outperform zero-shot LLMs on real-time termination prediction, while next-action prediction remains only moderately successful even for strong LLMs. Taken together, these results show that current models can capture some scam-related cues, yet still struggle to track how risk escalates and how manipulation unfolds across turns.

cs.CL↗

arXiv2026

From UI to Code: Mobile Ads Detection via LLM-Unified Static-Dynamic Analysis

Mobile advertisements (ads) are essential to the app economy, yet detecting them is challenging because ad content is dynamically fetched from remote servers and rendered through diverse user interfaces (UIs), making ads difficult to locate and trigger at runtime. To address this challenge, we present ADWISE, a novel framework that formulates mobile ads detection as LLM-guided, ad-oriented UI exploration. ADWISE first performs static program analysis to identify UI widgets used to place ads, which we call ad widgets. It then uses a grounded LLM reasoning loop to navigate toward and trigger these widgets under three complementary domain guidance signals: (1) WTG-based guidance, which provides global transition priors from a statically constructed window transition graph (WTG); (2) semantic guidance, which reasons over app functionality to prioritize user-likely interaction paths; and (3) structural guidance, which applies retrieval-augmented generation to match the current UI against recurring ad-heavy layouts from a knowledge base. By combining static program analysis with LLM-based reasoning over UI structure, app semantics, and retrieved analogies, ADWISE enables more effective ads detection in complex mobile UIs. Experiments on 100 benchmark apps show that ADWISE outperforms state-of-the-art baselines by 25.60% in ad widget detection. In addition, ADWISE uncovers 34.34% more ad regulation violations across six categories, directly benefiting downstream ad regulation.

cs.SE↗

arXiv2026

Agent+P: Guiding UI Agents via Symbolic Planning

Large Language Model (LLM)-based UI agents show great promise for UI automation but often hallucinate in long-horizon tasks due to their lack of understanding of the global UI transition structure. To address this, we introduce AGENT+P, a novel framework that leverages symbolic planning to guide LLM-based UI agents. Specifically, we model an app's UI transition structure as a UI Transition Graph (UTG), which allows us to reformulate the UI automation task as a pathfinding problem on the UTG. This further enables an off-the-shelf symbolic planner to generate a provably correct and optimal high-level plan, preventing the agent from redundant exploration and guiding the agent to achieve the automation goals. AGENT+P is designed as a plug-and-play framework to enhance existing UI agents. Evaluation on the AndroidWorld benchmark demonstrates that AGENT+P improves the success rates of state-of-the-art UI agents by up to 14.31% and reduces the action steps by 37.70%.

cs.MA↗

arXiv2025

ProGQL: A Provenance Graph Query System for Cyber Attack Investigation

Provenance analysis (PA) has recently emerged as an important solution for cyber attack investigation. PA leverages system monitoring to monitor system activities as a series of system audit events and organizes these events as a provenance graph to show the dependencies among system activities, which can reveal steps of cyber attacks. Despite their potential, existing PA techniques face two critical challenges: (1) they are inflexible and non-extensible, making it difficult to incorporate analyst expertise, and (2) they are memory inefficient, often requiring>100GB of RAM to hold entire event streams, which fundamentally limits scalability and deployment in real-world environments. To address these limitations, we propose the ProGQL framework, which provides a domain-specific graph search language with a well-engineered query engine, allowing PA over system audit events and expert knowledge to be jointly expressed as a graph search query and thereby facilitating the investigation of complex cyberattacks. In particular, to support dependency searches from a starting edge required in PA, ProGQL introduces new language constructs for constrained graph traversal, edge weight computation, value propagation along weighted edges, and graph merging to integrate multiple searches. Moreover, the ProGQL query engine is optimized for efficient incremental graph search across heterogeneous database backends, eliminating the need for full in-memory materialization and reducing memory overhead. Our evaluations on real attacks demonstrate the effectiveness of the ProGQL language in expressing a diverse set of complex attacks compared with the state-of-the-art graph query language Cypher, and the comparison with the SOTA PA technique DEPIMPACT further demonstrates the significant improvement of the scalability brought by our ProGQL framework's design.

cs.CR↗

arXiv2025

BinCtx: Multi-Modal Representation Learning for Robust Android App Behavior Detection

Mobile app markets host millions of apps, yet undesired behaviors (e.g., disruptive ads, illegal redirection, payment deception) remain hard to catch because they often do not rely on permission-protected APIs and can be easily camouflaged via UI or metadata edits. We present BINCTX, a learning approach that builds multi-modal representations of an app from (i) a global bytecode-as-image view that captures code-level semantics and family-style patterns, (ii) a contextual view (manifested actions, components, declared permissions, URL/IP constants) indicating how behaviors are triggered, and (iii) a third-party-library usage view summarizing invocation frequencies along inter-component call paths. The three views are embedded and fused to train a contextual-aware classifier. On real-world malware and benign apps, BINCTX attains a macro F1 of 94.73%, outperforming strong baselines by at least 14.92%. It remains robust under commercial obfuscation (F1 84% post-obfuscation) and is more resistant to adversarial samples than state-of-the-art bytecode-only systems.

cs.CR↗

arXiv2025

PsyScam: A Benchmark for Psychological Techniques in Real-World Scams

Over the years, online scams have grown dramatically, with nearly 50% of global consumers encountering scam attempts each week. These scams cause not only significant financial losses to individuals and businesses, but also lasting psychological trauma, largely due to scammers' strategic employment of psychological techniques (PTs) to manipulate victims. Meanwhile, scammers continually evolve their tactics by leveraging advances in Large Language Models (LLMs) to generate diverse scam variants that easily bypass existing defenses. To address this pressing problem, we introduce PsyScam, a benchmark designed to systematically capture the PTs employed in real-world scam reports, and investigate how LLMs can be utilized to generate variants of scams based on the PTs and the contexts provided by these scams. Specifically, we collect a wide range of scam reports and ground its annotations of employed PTs in well-established cognitive and psychological theories. We further demonstrate LLMs' capabilities in generating through two downstream tasks: scam completion, and scam augmentation. Experimental results show that PsyScam presents significant challenges to existing models in both detecting and generating scam content based on the PTs used by real-world scammers. Our code and dataset are available.

cs.CR↗

arXiv2025

Privacy and Accuracy-Aware AI/ML Model Deduplication

With the growing adoption of privacy-preserving machine learning algorithms, such as Differentially Private Stochastic Gradient Descent (DP-SGD), training or fine-tuning models on private datasets has become increasingly prevalent. This shift has led to the need for models offering varying privacy guarantees and utility levels to satisfy diverse user requirements. However, managing numerous versions of large models introduces significant operational challenges, including increased inference latency, higher resource consumption, and elevated costs. Model deduplication is a technique widely used by many model serving and database systems to support high-performance and low-cost inference queries and model diagnosis queries. However, none of the existing model deduplication works has considered privacy, leading to unbounded aggregation of privacy costs for certain deduplicated models and inefficiencies when applied to deduplicate DP-trained models. We formalize the problems of deduplicating DP-trained models for the first time and propose a novel privacy- and accuracy-aware deduplication mechanism to address the problems. We developed a greedy strategy to select and assign base models to target models to minimize storage and privacy costs. When deduplicating a target model, we dynamically schedule accuracy validations and apply the Sparse Vector Technique to reduce the privacy costs associated with private validation data. Compared to baselines that do not provide privacy guarantees, our approach improved the compression ratio by up to $35\times$ for individual models (including large language models and vision transformers). We also observed up to $43\times$ inference speedup due to the reduction of I/O operations.

cs.CR↗

arXiv2025

Careful About What App Promotion Ads Recommend! Detecting and Explaining Malware Promotion via App Promotion Graph

In Android apps, their developers frequently place app promotion ads, namely advertisements to promote other apps. Unfortunately, the inadequate vetting of ad content allows malicious developers to exploit app promotion ads as a new distribution channel for malware. To help detect malware distributed via app promotion ads, in this paper, we propose a novel approach, named ADGPE, that synergistically integrates app user interface (UI) exploration with graph learning to automatically collect app promotion ads, detect malware promoted by these ads, and explain the promotion mechanisms employed by the detected malware. Our evaluation on 18, 627 app promotion ads demonstrates the substantial risks in the app promotion ecosystem.

cs.CR↗

arXiv2023

NODLINK: An Online System for Fine-Grained APT Attack Detection and Investigation

Advanced Persistent Threats (APT) attacks have plagued modern enterprises, causing significant financial losses. To counter these attacks, researchers propose techniques that capture the complex and stealthy scenarios of APT attacks by using provenance graphs to model system entities and their dependencies. Particularly, to accelerate attack detection and reduce financial losses, online provenance-based detection systems that detect and investigate APT attacks under the constraints of timeliness and limited resources are in dire need. Unfortunately, existing online systems usually sacrifice detection granularity to reduce computational complexity and produce provenance graphs with more than 100,000 nodes, posing challenges for security admins to interpret the detection results. In this paper, we design and implement NodLink, the first online detection system that maintains high detection accuracy without sacrificing detection granularity. Our insight is that the APT attack detection process in online provenance-based detection systems can be modeled as a Steiner Tree Problem (STP), which has efficient online approximation algorithms that recover concise attack-related provenance graphs with a theoretically bounded error. To utilize STP approximation algorithm frameworks for APT attack detection, we propose a novel design of in-memory cache, an efficient attack screening method, and a new STP approximation algorithm that is more efficient than the conventional one in APT attack detection while maintaining the same complexity. We evaluate NodLink in a production environment. The open-world experiment shows that NodLink outperforms two state-of-the-art (SOTA) online provenance analysis systems by achieving magnitudes higher detection and investigation accuracy while having the same or higher throughput.

cs.CR↗

arXiv2023

Are we there yet? An Industrial Viewpoint on Provenance-based Endpoint Detection and Response Tools

Provenance-Based Endpoint Detection and Response (P-EDR) systems are deemed crucial for future APT defenses. Despite the fact that numerous new techniques to improve P-EDR systems have been proposed in academia, it is still unclear whether the industry will adopt P-EDR systems and what improvements the industry desires for P-EDR systems. To this end, we conduct the first set of systematic studies on the effectiveness and the limitations of P-EDR systems. Our study consists of four components: a one-to-one interview, an online questionnaire study, a survey of the relevant literature, and a systematic measurement study. Our research indicates that all industry experts consider P-EDR systems to be more effective than conventional Endpoint Detection and Response (EDR) systems. However, industry experts are concerned about the operating cost of P-EDR systems. In addition, our research reveals three significant gaps between academia and industry: (1) overlooking client-side overhead; (2) imbalanced alarm triage cost and interpretation cost; and (3) excessive server-side memory consumption. This paper's findings provide objective data on the effectiveness of P-EDR systems and how much improvements are needed to adopt P-EDR systems in industry.

cs.CR↗

arXiv2023

On the Security Risks of Knowledge Graph Reasoning

Knowledge graph reasoning (KGR) -- answering complex logical queries over large knowledge graphs -- represents an important artificial intelligence task, entailing a range of applications (e.g., cyber threat hunting). However, despite its surging popularity, the potential security risks of KGR are largely unexplored, which is concerning, given the increasing use of such capability in security-critical domains. This work represents a solid initial step towards bridging the striking gap. We systematize the security threats to KGR according to the adversary's objectives, knowledge, and attack vectors. Further, we present ROAR, a new class of attacks that instantiate a variety of such threats. Through empirical evaluation in representative use cases (e.g., medical decision support, cyber threat hunting, and commonsense reasoning), we demonstrate that ROAR is highly effective to mislead KGR to suggest pre-defined answers for target queries, yet with negligible impact on non-target ones. Finally, we explore potential countermeasures against ROAR, including filtering of potentially poisoning knowledge and training with adversarially augmented queries, which leads to several promising research directions.

cs.CR↗

arXiv2021

Trade or Trick? Detecting and Characterizing Scam Tokens on Uniswap Decentralized Exchange

The prosperity of the cryptocurrency ecosystem drives the need for digital asset trading platforms. Beyond centralized exchanges (CEXs), decentralized exchanges (DEXs) are introduced to allow users to trade cryptocurrency without transferring the custody of their digital assets to the middlemen, thus eliminating the security and privacy issues of traditional CEX. Uniswap, as the most prominent cryptocurrency DEX, is continuing to attract scammers, with fraudulent cryptocurrencies flooding in the ecosystem. In this paper, we take the first step to detect and characterize scam tokens on Uniswap. We first collect all the transactions related to Uniswap V2 exchange and investigate the landscape of cryptocurrency trading on Uniswap from different perspectives. Then, we propose an accurate approach for flagging scam tokens on Uniswap based on a guilt-by-association heuristic and a machine-learning powered technique. We have identified over 10K scam tokens listed on Uniswap, which suggests that roughly 50% of the tokens listed on Uniswap are scam tokens. All the scam tokens and liquidity pools are created specialized for the "rug pull" scams, and some scam tokens have embedded tricks and backdoors in the smart contracts. We further observe that thousands of collusion addresses help carry out the scams in league with the scam token/pool creators. The scammers have gained a profit of at least \$16 million from 39,762 potential victims. Our observations in this paper suggest the urgency to identify and stop scams in the decentralized finance ecosystem, and our approach can act as a whistleblower that identifies scam tokens at their early stages.

cs.CR↗

arXiv2021

Towards Robust Reasoning over Knowledge Graphs

Answering complex logical queries over large-scale knowledge graphs (KGs) represents an important artificial intelligence task, entailing a range of applications. Recently, knowledge representation learning (KRL) has emerged as the state-of-the-art approach, wherein KG entities and the query are embedded into a latent space such that entities that answer the query are embedded close to the query. Yet, despite its surging popularity, the potential security risks of KRL are largely unexplored, which is concerning, given the increasing use of such capabilities in security-critical domains (e.g., cyber-security and healthcare). This work represents a solid initial step towards bridging this gap. We systematize the potential security threats to KRL according to the underlying attack vectors (e.g., knowledge poisoning and query perturbation) and the adversary's background knowledge. More importantly, we present ROAR(Reasoning Over Adversarial Representations), a new class of attacks that instantiate a variety of such threats. We demonstrate the practicality of ROAR in two representative use cases (i.e., cyber-threat hunting and drug repurposing). For instance, ROAR attains over 99% attack success rate in misleading the threat intelligence engine to give pre-defined answers for target queries, yet without any impact on non-target ones. Further, we discuss potential countermeasures against ROAR, including filtering of poisoning facts and robust training with adversarial queries, which leads to several promising research directions.

cs.CR↗

arXiv2021

CHAMP: Characterizing Undesired App Behaviors from User Comments based on Market Policies

Millions of mobile apps have been available through various app markets. Although most app markets have enforced a number of automated or even manual mechanisms to vet each app before it is released to the market, thousands of low-quality apps still exist in different markets, some of which violate the explicitly specified market policies.In order to identify these violations accurately and timely, we resort to user comments, which can form an immediate feedback for app market maintainers, to identify undesired behaviors that violate market policies, including security-related user concerns. Specifically, we present the first large-scale study to detect and characterize the correlations between user comments and market policies. First, we propose CHAMP, an approach that adopts text mining and natural language processing (NLP) techniques to extract semantic rules through a semi-automated process, and classifies comments into 26 pre-defined types of undesired behaviors that violate market policies. Our evaluation on real-world user comments shows that it achieves both high precision and recall ($>0.9$) in classifying comments for undesired behaviors. Then, we curate a large-scale comment dataset (over 3 million user comments) from apps in Google Play and 8 popular alternative Android app markets, and apply CHAMP to understand the characteristics of undesired behavior comments in the wild. The results confirm our speculation that user comments can be used to pinpoint suspicious apps that violate policies declared by app markets. The study also reveals that policy violations are widespread in many app markets despite their extensive vetting efforts. CHAMP can be a \textit{whistle blower} that assigns policy-violation scores and identifies most informative comments for apps.

cs.SE↗

arXiv2021

Enabling Efficient Cyber Threat Hunting With Cyber Threat Intelligence

Log-based cyber threat hunting has emerged as an important solution to counter sophisticated attacks. However, existing approaches require non-trivial efforts of manual query construction and have overlooked the rich external threat knowledge provided by open-source Cyber Threat Intelligence (OSCTI). To bridge the gap, we propose ThreatRaptor, a system that facilitates threat hunting in computer systems using OSCTI. Built upon system auditing frameworks, ThreatRaptor provides (1) an unsupervised, light-weight, and accurate NLP pipeline that extracts structured threat behaviors from unstructured OSCTI text, (2) a concise and expressive domain-specific query language, TBQL, to hunt for malicious system activities, (3) a query synthesis mechanism that automatically synthesizes a TBQL query for hunting, and (4) an efficient query execution engine to search the big audit logging data. Evaluations on a broad set of attack cases demonstrate the accuracy and efficiency of ThreatRaptor in practical threat hunting.

cs.CR↗