Vulnerability Detection & Repair

Autonomous agent frameworks built upon large language models (LLMs) are evolving into complex, tool-integrated, and continuously operating systems, introducing security risks beyond traditional prompt-level vulnerabilities. As this paradigm is still at an early stage of development, a timely and systematic understanding of its security implications is increasingly important. Although a growing body of work has examined different attack surfaces and defense problems in agent systems, existing studies remain scattered across individual aspects of agent security, and there is still a lack of a layered review on this topic. To address this gap, this survey presents a layered review of security risks and defense strategies in autonomous agent frameworks, with OpenClaw as a case study. We organize the analysis into four security-relevant layers: the context and instruction layer, the tool and action layer, the state and persistence layer, and the ecosystem and automation layer. For each layer, we summarize its functional role, representative security risks, and corresponding defense strategies. Based on this layered analysis, we further identify that threats in autonomous agent frameworks may propagate across layers, from manipulated inputs to unsafe actions, persistent state contamination, and broader ecosystem-level impact. Finally, we highlight potential key challenges, including research imbalance across layers, the lack of long-horizon evaluation, and weak ecosystem trust models, and outline future directions toward more systematic and integrated defenses.

Recent research has demonstrated the potential of Large Language Models (LLMs) for autonomous penetration testing, particularly when using cloud-based restricted-weight models. However, reliance on such models introduces security, privacy, and sovereignty concerns, motivating the use of locally hosted open-weight alternatives. Prior work shows that small open-weight models perform poorly on automated Linux privilege escalation, limiting their practical applicability. In this paper, we present a systematic empirical study of whether targeted system-level and prompting interventions can bridge this performance gap. We analyze failure modes of open-weight models in autonomous privilege escalation, map them to established enhancement techniques, and evaluate five concrete interventions (chain-of-thought prompting, retrieval-augmented generation, structured prompts, history compression, and reflective analysis) implemented as extensions to hackingBuddyGPT. Our results show that open-weight models can match or outperform cloud-based baselines such as GPT-4o. With our treatments enabled, Llama3.1 70B exploits 83% of tested vulnerabilities, while smaller models including Llama3.1 8B and Qwen2.5 7B achieve 67% when using guidance. A full-factorial ablation study over all treatment combinations reveals that reflection-based treatments contribute most, while also identifying vulnerability discovery as a remaining bottleneck for local models.

Alignment faking (AF) occurs when an LLM strategically complies with training objectives to avoid value modification, reverting to prior preferences once monitoring is lifted. Current detection methods focus on conversational settings and rely primarily on Chain-of-Thought (CoT) analysis, which provides a reliable signal when strategic reasoning surfaces, but cannot distinguish deception from capability failures if traces are absent or unfaithful. We formalize AF as a composite behavioural event and detect it through observable tool selection, where the LLM selects the safe tool when unmonitored, but switches to the unsafe tool under monitoring that rewards helpfulness over safety, while its reasoning still acknowledges the safe choice. We release a dataset of 108 enterprise IT scenarios spanning Security, Privacy, and Integrity domains under Corruption and Sabotage pressures. Evaluating six frontier LLMs across five independent runs, we find mean AF detection rates between 3.5% and 23.7%, with vulnerability profiles varying by domain and pressure type. These results suggest that susceptibility reflects training methodology rather than capability alone.

We present VulStyle, a multi-modal software vulnerability detection model that jointly encodes function-level source code, non-terminal Abstract Syntax Tree (AST) structure, and code stylometry (CStyle) features. Prior work in code representation primarily leverages token-level models or full AST trees, often missing stylistic cues indicative of risky programming practices, or incurring high structural overhead. Our approach selects only non-terminal AST nodes, reducing input complexity while preserving semantic hierarchy, and integrates syntactic and lexical CStyle features as auxiliary vulnerability signals. VulStyle is pre-trained using masked language modeling on 4.9M functions across seven programming languages, and fine-tuned across five benchmark datasets: Devign, BigVul, DiverseVul, REVEAL, and VulDeePecker. VulStyle achieves state-of-the-art performance on BigVul and VulDeePecker, improving F1 by 4-48% over strong transformer baselines, and attains competitive or best-average performance across all benchmarks. We contribute an ablation study isolating the effect of CStyle and AST structure, error case analysis, and a threat model situating the detection task in attacker-realistic scenarios.

Source code and its accompanying comments are complementary yet naturally aligned modalities-code encodes structural logic while comments capture developer intent. However, existing vulnerability detection methods mostly rely on single-modality code representations, overlooking the complementary semantic information embedded in comments and thus limiting their generalization across complex code structures and logical relationships. To address this, we propose MultiVul, a multimodal contrastive framework that aligns code and comment representations through dual similarity learning and consistency regularization, augmented with diverse code-text pairs to improve robustness. Experiments on widely adopted DiverseVul and Devign datasets across four large language models (LLMs) (i.e., DeepSeek-Coder-6.7B, Qwen2.5-Coder-7B, StarCoder2-7B, and CodeLlama-7B) show that MultiVul achieves up to 27.07% F1 improvement over prompting-based methods and 13.37% over code-only Fine-Tuning, while maintaining comparable inference efficiency.

Code understanding models increasingly rely on pretrained language models (PLMs) and graph neural networks (GNNs), which capture complementary semantic and structural information. We conduct a controlled empirical study of PLM-GNN hybrids for code classification and vulnerability detection tasks by systematically pairing three code-specialized PLMs with three foundational GNN architectures. We compare these hybrids against PLM-only and GNN-only baselines on Java250 and Devign, including an identifier-obfuscation setting. Across both tasks, hybrids consistently outperform GNN-only baselines and often improve ranking quality over frozen PLMs. On Devign, performance and robustness are more sensitive to the PLM feature source than to the GNN backbone. We also find that larger PLMs are not necessarily better feature extractors in this pipeline, and that the PLM choice has more impact than the GNN choice. Finally, we distill these findings into practical guidelines for PLM-GNN design choices in code classification and vulnerability detection.

Context: Software vulnerabilities pose significant security threats to software systems, especially as software is increasingly used across many areas of daily life, including health, government, and finance. Recently, transformer-based models have demonstrated promising results in automatic software vulnerability identification due to their robust contextual modelling and representation learning capabilities. Objectives: While numerous systematic literature reviews (SLRs) have examined machine learning and deep learning methods for identifying vulnerabilities, a more transformer-centric analysis remains to be explored. This SLR critically analysed 80 studies published between 2021 and 2025 that utilised transformer models to identify software vulnerabilities. Methods: Using Kitchenhams SLR guidelines, we methodically evaluate current research from various perspectives, encompassing study trends, datasets and sources, programming languages, transformer frameworks, detection detail levels, assessment metrics, reference models, types of vulnerabilities, and experimental configurations. Results: We classify transformer models into encoder, decoder, and combined architectures and analyse both pre-trained and fine-tuned versions utilized on source code, logs, and smart contracts. The results emphasise prevailing research trends, frequently utilised benchmarks, and main baselines. It also uncovers crucial technical issues like data imbalance, interpretability, scalability, and generalization across programming languages. Conclusion: By integrating current evidence and recognising unaddressed research areas, this SLR provides a consolidated resource for researchers and professionals seeking to develop more reliable, precise, and interpretable transformer-based vulnerability identification systems.

Large (vision-)language models exhibit remarkable capability but remain highly susceptible to jailbreaking. Existing safety training approaches aim to have the model learn a refusal boundary between safe and unsafe, based on the user's intent. It has been found that this binary training regime often leads to brittleness, since the user intent cannot reliably be evaluated, especially if the attacker obfuscates their intent, and also makes the system seem unhelpful. In response, frontier models, such as GPT-5, have shifted from refusal-based safeguards to safe completion, that aims to maximize helpfulness while obeying safety constraints. However, safe completion could be exploited when a user pretends their intention is benign. Specifically, this intent inversion would be effective in multi-turn conversation, where the attacker has multiple opportunities to reinforce their deceptively benign intent. In this work, we introduce a novel multi-turn jailbreaking method that exploits this vulnerability. Our approach gradually builds conversational trust by simulating benign-seeming intentions and by exploiting the consistency property of the model, ultimately guiding the target model toward harmful, detailed outputs. Most crucially, our approach also uncovered an additional class of model vulnerability that we call para-jailbreaking that has been unnoticed up to now. Para-jailbreaking describes the situation where the model may not reveal harmful direct reply to the attack query, however the information that it reveals is nevertheless harmful. Our contributions are threefold. First, it achieves high success rates against frontier models including GPT-5-thinking and Claude-Sonnet-4.5. Second, our approach revealed and addressed para-jailbreaking harmful output. Third, experiments on multimodal VLM models showed that our approach outperformed state-of-the-art models.

Multi-component natural language processing (NLP) pipelines are increasingly deployed for high-stakes decisions, yet no existing adversarial method can test their robustness under realistic conditions: binary-only feedback, no gradient access, and strict query budgets. We formalize this strict black-box threat model and propose a two-agent evasion framework operating in a semantic perturbation space. An Attacker Agent generates meaning-preserving rewrites while a Prompt Optimization Agent refines the attack strategy using only binary decision feedback within a 10-query budget. Evaluated against four evidence-based misinformation detection pipelines, the framework achieves evasion rates of 19.95 to 40.34% on modern large language model (LLM) based systems, compared to at most 3.90% for token-level perturbation baselines that rely on surrogate models because they cannot operate under our threat model. A legacy system relying on static lexical retrieval exhibits near-total vulnerability 97.02%, establishing a lower bound that exposes how architectural choices govern the attack surface. Evasion effectiveness is associated with three architectural properties: evidence retrieval mechanism, retrieval-inference coupling, and baseline classification accuracy. The iterative prompt optimization yields the largest marginal gains against the most robust targets, confirming that adaptive strategy discovery is essential when evasion is non-trivial. Analysis of successful rewrites reveals four exploitation patterns, each targeting failures at distinct pipeline stages. A pattern-informed defense reduces the evasion rate by up to 65.18%.

Accurate mapping between Common Vulnerabilities and Exposures (CVE) and Common Weakness Enumeration (CWE) entries is critical for effective vulnerability management and risk assessment. However, public databases, such as the National Vulnerability Database (NVD), suffer from inconsistent and incomplete CVE to CWE mappings, complicating automated analysis and remediation. We introduce FixV2W, a lightweight approach that leverages knowledge graph embeddings and longitudinal trends to improve mapping accuracy of the NVD. FixV2W systematically analyzes historical remapping patterns and leverages hierarchical relationships within NVD and CWE data to predict more precise CWE mappings for vulnerabilities linked to Prohibited or Discouraged categories. We run extensive experimental evaluation of FixV2W, based on test data set collected between August 2021 and December 2024. Considering the Top 10 ranked predictions, the results show that FixV2W predicts the correct CWE mappings for 69% of exploited vulnerabilities that had invalid CWEs before they were exploited. We also show that FixV2W significantly improves the performance of ML models relying on NVD data. For instance, for a model geared at uncovering unknown CVE-CWE mappings, FixV2W improves the Mean Reciprocal Rank (MRR) from 0.174 to 0.608. These results show that FixV2W is a promising approach to identify and thwart emerging threats.

Automated code vulnerability detection is critical for software security, yet existing approaches face a fundamental trade-off between detection accuracy and computational cost. We propose a heterogeneous multi-agent architecture inspired by game-theoretic principles, combining cloud-based LLM experts with a local lightweight verifier. Our "3+1" architecture deploys three cloud-based expert agents (DeepSeek-V3) that analyze code from complementary perspectives - code structure, security patterns, and debugging logic - in parallel, while a local verifier (Qwen3-8B) performs adversarial validation at zero marginal cost. We formalize this design through a two-layer game framework: (1) a cooperative game among experts capturing super-additive value from diverse perspectives, and (2) an adversarial verification game modeling quality assurance incentives. Experiments on 262 real samples from the NIST Juliet Test Suite across 14 CWE types, with balanced vulnerable and benign classes, demonstrate that our approach achieves a 77.2% F1 score with 62.9% precision and 100% recall at $0.002 per sample - outperforming both a single-expert LLM baseline (F1 71.4%) and Cppcheck static analysis (MCC 0). The adversarial verifier significantly improves precision (+10.3 percentage points, p < 1e-6, McNemar's test) by filtering false positives, while parallel execution achieves a 3.0x speedup. Our work demonstrates that game-theoretic design principles can guide effective heterogeneous multi-agent architectures for cost-sensitive software engineering tasks.

As artificial intelligence (AI) systems are increasingly deployed across critical domains, their security vulnerabilities pose growing risks of high-profile exploits and consequential system failures. Yet systematic approaches to evaluating AI security remain underdeveloped. In this paper, we introduce AVISE (AI Vulnerability Identification and Security Evaluation), a modular open-source framework for identifying vulnerabilities in and evaluating the security of AI systems and models. As a demonstration of the framework, we extend the theory-of-mind-based multi-turn Red Queen attack into an Adversarial Language Model (ALM) augmented attack and develop an automated Security Evaluation Test (SET) for discovering jailbreak vulnerabilities in language models. The SET comprises 25 test cases and an Evaluation Language Model (ELM) that determines whether each test case was able to jailbreak the target model, achieving 92% accuracy, an F1-score of 0.91, and a Matthews correlation coefficient of 0.83. We evaluate nine recently released language models of diverse sizes with the SET and find that all are vulnerable to the augmented Red Queen attack to varying degrees. AVISE provides researchers and industry practitioners with an extensible foundation for developing and deploying automated SETs, offering a concrete step toward more rigorous and reproducible AI security evaluation.

The rapidly evolving Node$.$js ecosystem currently includes millions of packages and is a critical part of modern software supply chains, making vulnerability detection of Node$.$js packages increasingly important. However, traditional program analysis struggles in this setting because of dynamic JavaScript features and the large number of package dependencies. Recent advances in large language models (LLMs) and the emerging paradigm of LLM-based agents offer an alternative to handcrafted program models. This raises the question of whether an LLM-centric, tool-augmented approach can effectively detect and confirm taint-style vulnerabilities (e.g., arbitrary command injection) in Node$.$js packages. We implement LLMVD$.$js, a multi-stage agent pipeline to scan code, propose vulnerabilities, generate proof-of-concept exploits, and validate them through lightweight execution oracles; and systematically evaluate its effectiveness in taint-style vulnerability detection and confirmation in Node$.$js packages without dedicated static/dynamic analysis engines for path derivation. For packages from public benchmarks, LLMVD$.$js confirms 84% of the vulnerabilities, compared to less than 22% for prior program analysis tools. It also outperforms a prior LLM-program-analysis hybrid approach while requiring neither vulnerability annotations nor prior vulnerability reports. When evaluated on a set of 260 recently released packages (without vulnerability groundtruth information), traditional tools produce validated exploits for few ($\leq 2$) packages, while LLMVD$.$js generates validated exploits for 36 packages.

Cross-site scripting (XSS) remains a persistent web security vulnerability, especially because obfuscation can change the surface form of a malicious payload while preserving its behavior. These transformations make it difficult for traditional and machine learning-based detection systems to reliably identify attacks. Existing approaches for generating obfuscated payloads often emphasize syntactic diversity, but they do not always ensure that the generated samples remain behaviorally valid. This paper presents a structured pipeline for generating and evaluating obfuscated XSS payloads using large language models (LLMs). The pipeline combines deterministic transformation techniques with LLM-based generation and uses a browser- based runtime evaluation procedure to compare payload behavior in a controlled execution environment. This allows generated samples to be assessed through observable runtime behavior rather than syntactic similarity alone. In the evaluation, an untuned baseline language model achieves a runtime behavior match rate of 0.15, while fine-tuning on behavior-preserving source-target obfuscation pairs improves the match rate to 0.22. Although this represents a measurable improvement, the results show that current LLMs still struggle to generate obfuscations that preserve observed runtime behavior. A downstream classifier evaluation further shows that adding generated payloads does not improve detection performance in this setting, although behavior- filtered generated samples can be incorporated without materially degrading performance. Overall, the study demonstrates both the promise and the limits of applying generative models to adversarial security data generation and emphasizes the importance of runtime behavior checks in improving the quality of generated data for downstream detection systems.

LLM-assisted defect discovery has a precision crisis: plausible-but-wrong reports overwhelm maintainers and degrade credibility for real findings. We present Refute-or-Promote, an inference-time reliability pattern combining Stratified Context Hunting (SCH) for candidate generation, adversarial kill mandates, context asymmetry, and a Cross-Model Critic (CMC). Adversarial agents attempt to disprove candidates at each promotion gate; cold-start reviewers are intended to reduce anchoring cascades; cross-family review can catch correlated blind spots that same-family review misses. Over a 31-day campaign across 7 targets (security libraries, the ISO C++ standard, major compilers), the pipeline killed roughly 79% of 171 candidates before advancing to disclosure (retrospective aggregate); on a consolidated-protocol subset (lcms2, wolfSSL; n=30), the prospective kill rate was 83%. Outcomes: 4 CVEs (3 public, 1 embargoed); LWG 4549 accepted to the C++ working paper; 5 merged C++ editorial PRs; 3 compiler conformance bugs; 8 merged security-related fixes without CVE; an RFC 9000 errata filed under committee review; and 1+ FIPS 140-3 normative compliance issues under coordinated disclosure -- all evaluated by external acceptance, not benchmarks. The most instructive failure: ten dedicated reviewers unanimously endorsed a non-existent Bleichenbacher padding oracle in OpenSSL's CMS module; it was killed only by a single empirical test, motivating the mandatory empirical gate. No vulnerability was discovered autonomously; the contribution is external structure that filters LLM agents' persistent false positives. As a preliminary transfer test beyond defect discovery, a simplified cross-family critique variant also solved five previously unsolved SymPy instances on SWE-bench Verified and one SWE-rebench hard task.

Detecting jailbreak behaviour in large language models remains challenging, particularly when strongly aligned models produce harmful outputs only rarely. In this work, we present an empirical study of output based jailbreak detection under realistic conditions using the JailbreakBench Behaviors dataset and multiple generator models with varying alignment strengths. We evaluate both a lexical TF-IDF detector and a generation inconsistency based detector across different sampling budgets. Our results show that single output evaluation systematically underestimates jailbreak vulnerability, as increasing the number of sampled generations reveals additional harmful behaviour. The most significant improvements occur when moving from a single generation to moderate sampling, while larger sampling budgets yield diminishing returns. Cross generator experiments demonstrate that detection signals partially generalise across models, with stronger transfer observed within related model families. A category level analysis further reveals that lexical detectors capture a mixture of behavioural signals and topic specific cues, rather than purely harmful behaviour. Overall, our findings suggest that moderate multi sample auditing provides a more reliable and practical approach for estimating model vulnerability and improving jailbreak detection in large language models. Code will be released.

Agentic security systems increasingly audit live targets with tool-using LLMs, but prior systems fix a single coordination topology, leaving unclear when additional agents help and when they only add cost. We treat topology choice as an empirical systems question. We introduce a controlled benchmark of 20 interactive targets (10 web/API and 10 binary), each exposing one endpoint-reachable ground-truth vulnerability, evaluated in whitebox and blackbox modes. The core study executes 600 runs over five architecture families, three model families, and both access modes, with a separate 60-run long-context pilot reported only in the appendix. On the completed core benchmark, detection-any reaches 58.0% and validated detection reaches 49.8%. MAS-Indep attains the highest validated detection rate (64.2%), while SAS is the strongest efficiency baseline at $0.058 per validated finding. Whitebox materially outperforms blackbox (67.0% vs. 32.7% validated detection), and web materially outperforms binary (74.3% vs. 25.3%). Bootstrap confidence intervals and paired target-level deltas show that the dominant effects are observability and domain, while some leading whitebox topologies remain statistically close. The main result is a non-monotonic cost-quality frontier: broader coordination can improve coverage, but it does not dominate once latency, token cost, and exploit-validation difficulty are taken into account.

While attack graphs are useful for identifying major cybersecurity threats affecting a system, they do not provide operational support for determining the likelihood of having a known vulnerability exploited, or that critical system nodes are likely to be compromised. In this paper, we perform dynamic risk assessment by combining Bayesian Attack Graphs (BAGs) and online monitoring of system behavior through process mining. Specifically, the proposed approach applies process mining techniques to characterize malicious network traffic and derive evidence regarding the probability of having a vulnerability actively exploited. This evidence is then provided to a BAG, which updates its conditional probability tables accordingly, enabling dynamic assessment of vulnerability exploitation. We apply our method to a cybersecurity testbed instantiating several machines deployed on different subnets and affected by several CVE vulnerabilities. The testbed is stimulated with both benign traffic and malicious behavior, which simulates network attack patterns aimed at exploiting the CVE vulnerabilities. The results indicate that our proposal effectively detects whether vulnerabilities are being actively exploited, allowing for an updated assessment of the probability of system compromise.

Large Language Models (LLMs) have demonstrated remarkable capabilities across various cybersecurity tasks, including vulnerability classification, detection, and patching. However, their potential in automated vulnerability report documentation and analysis remains underexplored. We present RAVEN (Retrieval Augmented Vulnerability Exploration Network), a framework leveraging LLM agents and Retrieval Augmented Generation (RAG) to synthesize comprehensive vulnerability analysis reports. Given vulnerable source code, RAVEN generates reports following the Google Project Zero Root Cause Analysis template. The framework uses four modules: an Explorer agent for vulnerability identification, a RAG engine retrieving relevant knowledge from curated databases including Google Project Zero reports and CWE entries, an Analyst agent for impact and exploitation assessment, and a Reporter agent for structured report generation. To ensure quality, RAVEN includes a task specific LLM Judge evaluating reports across structural integrity, ground truth alignment, code reasoning quality, and remediation quality. We evaluate RAVEN on 105 vulnerable code samples covering 15 CWE types from the NIST-SARD dataset. Results show an average quality score of 54.21%, supporting the effectiveness of our approach for automated vulnerability documentation.

Assessing the security posture of modern computing systems typically requires the use of multiple specialized tools. These tools focus on different aspects such as configuration compliance, file integrity, and vulnerability exposure, and their outputs are often difficult to interpret collectively. This paper introduces the Unified Compliance Aggregator (UCA), a framework that integrates several open-source security tools into a single composite score representing overall system security. The proposed framework combines outputs from Lynis, OpenSCAP (STIG and CIS profiles), AIDE, Tripwire, and Nmap NSE. A normalization process converts heterogeneous outputs into a consistent 0 to 100 scale, followed by weighted aggregation. We also introduce a logarithmic scoring model for file integrity measurements to address limitations observed in prior linear approaches. Experiments were conducted on Ubuntu 22.04 across different hardening levels and environments. Results show consistent improvement in composite scores as systems are hardened, while also revealing contrasting behavior between compliance and file integrity tools. Two case studies, a basic web server and a DVWA-based system illustrate how the framework can be applied in practical scenarios.

Rowhammer on GPU DRAM has enabled adversarial bit flips in model weights; shared KV-cache blocks in LLM serving systems present an analogous but previously unexamined target. In vLLM's Prefix Caching, these blocks exist as a single physical copy without integrity protection. Using software fault injection under ideal bit targeting, we characterize worst-case severity and identify three properties: (1) Silent divergence - 13 of 16 BF16 bit positions produce coherent but altered outputs, indistinguishable from legitimate responses without a clean baseline. (2) Selective propagation - only requests sharing the targeted prefix are affected. (3) Persistent accumulation - no temporal decay occurs, so cumulative damage grows linearly with subsequent requests. Together, these constitute a threat profile distinct from weight corruption: silent divergence and selective propagation enable detection evasion; persistent accumulation then proceeds unchecked, yielding damage amplification bounded only by how long the block remains cached. A checksum-based countermeasure detects any single-bit corruption at scheduling time, bounding cumulative damage to one batch independent of the block's cache lifetime, with negligible overhead. These results argue for integrity protection of prefix blocks before end-to-end exploitation is demonstrated.

Large Language Models (LLMs) can generate plausible code patches, but plausibility is not enough for automated repair: a patch must compile, pass tests, and remove the target vulnerability. We present SynthFix, a neuro-symbolic repair framework that combines supervised repair learning with compiler-informed feedback. During training, a lightweight router selects between Supervised Fine-Tuning (SFT) for common repair patterns and Reward Fine-Tuning (RFT) for examples that benefit from symbolic feedback. The reward combines static structure, lint/compile checks, security scanning, and public execution tests where available; at inference time, the same evidence guides best-of-K candidate selection under a greedy floor. Across five code LLMs (1.3B-7B) on pyrepair, CodeFlaws, and SVEN, SynthFix improves deployable repair metrics over SFT-only and RFT-only baselines, with relative gains up to 54 percent in functional correctness and 14 percent in security clearance. Our code and data are available at https://github.com/CoderDoge1108/SynthFix.

Large language models write production code, and yet they routinely introduce well-known vulnerabilities. We show that this is not a knowledge deficit: the same models that generate insecure code, correctly identify and explain the vulnerability when asked directly, this is a gap we call the Format-Reliability Gap. Mechanistic analysis reveals the cause: security representations are encoded from the earliest layers but remain computationally inert until the final layer, where format-compliance demands compete with them. Because the failure is localized to a single layer, per-vulnerability steering vectors reduce insecure generation by up to 74% with negligible overhead. The mechanism and the fix generalize across five models, three architecture families, and six vulnerability types, suggesting insecure code generation is an interpretability problem, not a training artifact.

Text-to-image (T2I) models are increasingly optimized for following user instructions faithfully. However, we find that this capability introduces a safety vulnerability we call Multi-Concept Compositional Unsafety (MCCU). MCCU occurs when multiple individually safe concepts, if combined in a single generation request, lead to harmful or sensitive visual outputs. Unlike prior jailbreak settings, MCCU does not rely on adversarial prompts, model access, or explicitly disallowed content. Instead, the risk emerges from how the model composes multiple safe visual concepts into a single scene. To systematically measure this threat, we build TwoHamsters, a large-scale evaluation framework consisting of 20k prompts, 51 curated concept pairs, and six risk categories. We evaluate 13 T2I models under a black-box setting. Our results show a clear conflict between instruction-following and safety: models that follow prompts more faithfully tend to produce more MCCU failures. For example, FLUX.1 achieves a 99.35% Unsafe Alignment Rate while only reaching a 1.57% MCCU Defense Rate. We further evaluate three representative defenses, including safety filtering, MCCU-specific detector fine-tuning, and concept erasure, all of which fail against unseen concept combinations. Our findings suggest that compositional reasoning in T2I models creates an attack surface that is not captured by existing safety mechanisms. We anticipate the release of TwoHamsters will catalyze community development of advanced generative defense mechanisms.

Software supply chain security compromises often stem from cascaded interactions of vulnerabilities, for example, between multiple vulnerable components. Yet, Software Bill of Materials (SBOM)-based pipelines for security analysis typically treat scanner findings as independent per-CVE (Common Vulnerabilities and Exposures) records. We propose a new research direction based on learning multi-vulnerability attack chains through a novel SBOM-driven graph-learning approach. This treats SBOM structure and scanner outputs as a dependency-constrained evidence graph rather than a flat list of vulnerabilities. We represent vulnerability-enriched CycloneDX SBOMs as heterogeneous graphs whose nodes capture software components and known vulnerabilities (i.e, CVEs), connected by typed relations, such as dependency and vulnerability links. We train a Heterogeneous Graph Attention Network (HGAT) to predict whether a component is associated with at least one known vulnerability as a feasibility check for learning over this structure. Additionally, we frame the discovery of cascading vulnerabilities as CVE-pair link prediction using a lightweight Multi-Layer Perceptron (MLP) neural network trained on documented multi-vulnerability chains. Validated on 200 real-world SBOMs from the Wild SBOMs public dataset, the HGAT component classifier achieves 91.03% Accuracy and 74.02% F1-score, while the cascade predictor model (MLP) achieves a Receiver Operating Characteristic - Area Under Curve (ROC-AUC) of 0.93 on a seed set of 35 documented attack chains.

Smart contracts govern billions of dollars in decentralized finance (DeFi), yet automated vulnerability detection remains challenging because many vulnerabilities are tightly coupled with project-specific business logic. We observe that recurring vulnerabilities across diverse DeFi business models often share the same underlying economic mechanisms, which we term DeFi semantics, and that capturing these shared abstractions can enable more systematic auditing. Building on this insight, we propose Knowdit, a knowledge-driven, agentic workflow for smart contract vulnerability detection. Knowdit first constructs an auditing knowledge graph from historical human audit reports, linking fine-grained DeFi semantics with recurring vulnerability patterns. Given a new project, a multi-agent pipeline leverages this knowledge through an iterative loop of specification generation, Proof-of-Concept (PoC) synthesis, PoC execution, and finding reflection, driven by a shared repository index. We evaluate Knowdit on 11 recent Code4rena projects with 84 ground-truth vulnerabilities. Knowdit detects all 21 high-severity and 90% of medium-severity vulnerabilities without false positives, fully covering eight projects, significantly outperforming all baselines. Applied to seven real-world projects, Knowdit further discovers 9 high- and 36 medium-severity previously unknown vulnerabilities, securing millions in liquidity and proving its outstanding performance.

Adversarial attacks can readily compromise deep neural networks (DNNs). In particular, transferable attacks (TAs) exploit the shared vulnerabilities among DNNs, enabling perturbations crafted on surrogates to transfer to unseen models. Training-time and post-attack defenses have been extensively studied for combating TAs. Orthogonal to these approaches, preemptive robustification (PR) has emerged as a pre-attack defense that enhances the robustness of benign samples by superimposing protective variations before attacks. Despite its promise, PR remains underexplored and faces several important limitations. First, dependence on well-trained surrogate classifiers limits applicability, as surrogates are task-specific and may even be unavailable in some practical settings. Second, the required iterative optimization or dedicated PR generator training incurs substantial costs. Third, the generated variations are opaque to humans. To address these, we seek an efficient PR that is surrogate-free, optimization-free, training-free, and human-interpretable. Intriguingly, we discover a numerical correlation between the shared vulnerabilities of DNNs and Laplacian responses, with their cosine similarity being significantly negative. This indicates that negated high-frequency response constitutes an important component of shared vulnerabilities. Consequently, strengthening Laplacian responses counteracts this component, improving resistance to TAs. Building upon this insight, we propose Fast Preemptive Robustification (FPR), which performs Laplacian sharpening via a single channel-wise convolution with a 3\times3 kernel. FPR is simple yet effective, as demonstrated by extensive experiments. Specifically, FPR reduces the attack success rate (ASR) of untargeted TAs by 12.7% and that of targeted TAs from 10.7% to 4.1%. Code will be released publicly.

WebAssembly's (Wasm) monolithic linear memory turns a single memory-corruption bug into a bidirectional threat: a compromised module can attack its embedding host, and a malicious host can tamper with a trusted module's state. Existing defenses require custom runtimes or source changes, and none can verify runtime integrity under an adversarial host. We present Walma, a framework for WebAssembly linear memory attestation that renders snapshots of linear memory as grayscale images and classifies them with a convolutional neural network. The network reads the row-aligned structure that compiled code imposes on memory and detects corruption that byte and texture statistics miss. On structured targets this extends to tampering that no program input triggers, such as direct memory writes by a malicious host. On real-world CVE-affected applications, hiding corruption from Walma's verdict requires overwriting a large fraction of memory, hundreds of kilobytes to megabytes on our targets. A Shannon-entropy analysis of benign memory bounds the class of out-of-band edits that the learned prior can detect. Across the 53 binaries of the WABench suite, attestation costs 1.07x--1.69x at the host boundary, making continuous, state-centric memory attestation for Wasm practical.

Frontier LLM companies have repeatedly assured courts and regulators that their models do not store copies of training data. They further rely on safety alignment strategies via RLHF, system prompts, and output filters to block verbatim regurgitation of copyrighted works, and have cited the efficacy of these measures in their legal defenses against copyright infringement claims. We show that finetuning bypasses these protections: by training models to expand plot summaries into full text, a task naturally suited for commercial writing assistants, we cause GPT-4o, Gemini-2.5-Pro, and DeepSeek-V3.1 to reproduce up to 85-90% of held-out copyrighted books, with single verbatim spans exceeding 460 words, using only semantic descriptions as prompts and no actual book text. This extraction generalizes across authors: finetuning exclusively on Haruki Murakami's novels unlocks verbatim recall of copyrighted books from over 30 unrelated authors. The effect is not specific to any training author or corpus: random author pairs and public-domain finetuning data produce comparable extraction, while finetuning on synthetic text yields near-zero extraction, indicating that finetuning on individual authors' works reactivates latent memorization from pretraining. Three models from different providers memorize the same books in the same regions ($r \ge 0.90$), pointing to an industry-wide vulnerability. Our findings offer compelling evidence that model weights store copies of copyrighted works and that the security failures that manifest after finetuning on individual authors' works undermine a key premise of recent fair use rulings, where courts have conditioned favorable outcomes on the adequacy of measures preventing reproduction of protected expression.

While prior red-teaming efforts have focused on eliciting harmful text outputs from large language models (LLMs), such approaches fail to capture agent-specific vulnerabilities that emerge through multi-step tool execution, particularly in rapidly growing ecosystems such as the Model Context Protocol (MCP). To address this gap, we propose a trajectory-aware evolutionary search method, T-MAP, which leverages execution trajectories to guide the discovery of adversarial prompts. Our approach enables the automatic generation of attacks that not only bypass safety guardrails but also reliably realize harmful objectives through actual tool interactions. Empirical evaluations across diverse MCP environments demonstrate that T-MAP substantially outperforms baselines in attack realization rate (ARR) and remains effective against frontier models, including GPT-5.2, Gemini-3-Pro, Qwen3.5, and GLM-5, thereby revealing previously underexplored vulnerabilities in autonomous LLM agents. Code is available at https://github.com/pwnhyo/T-MAP.

Large language models (LLMs) increasingly act as autonomous agents, using tools to execute code, read and write files, and access networks, creating novel security risks. To mitigate these risks, agents are commonly deployed and evaluated in isolated "sandbox" environments, often implemented using Docker/OCI containers. We introduce SANDBOXESCAPEBENCH, an open benchmark that safely measures an LLM's capacity to break out of these sandboxes. The benchmark is implemented as an Inspect AI Capture the Flag (CTF) evaluation utilising a nested sandbox architecture with the outer layer containing the flag and no known vulnerabilities. Following a threat model of a motivated adversarial agent with shell access inside a container, SANDBOXESCAPEBENCH covers a spectrum of sandboxescape mechanisms spanning misconfiguration, privilege allocation mistakes, kernel flaws, and runtime/orchestration weaknesses. We find that, when vulnerabilities are added, LLMs are able to identify and exploit them, showing that use of evaluation like SANDBOXESCAPEBENCH is needed to ensure sandboxing continues to provide the encapsulation needed for highly-capable models.

Semantic caching has emerged as a pivotal technique for scaling LLM applications, widely adopted by major providers including AWS and Microsoft. By utilizing semantic embedding vectors as cache keys, this mechanism effectively minimizes latency and redundant computation for semantically similar queries. In this work, we conceptualize semantic cache keys as a form of fuzzy hashes. We demonstrate that the locality required to maximize cache hit rates fundamentally conflicts with the cryptographic avalanche effect necessary for collision resistance. Our conceptual analysis formalizes this inherent trade-off between performance (locality) and security (collision resilience), revealing that semantic caching is naturally vulnerable to key collision attacks. While prior research has focused on side-channel and privacy risks, we present the first systematic study of integrity risks arising from cache collisions. We introduce CacheAttack, an automated framework for launching black-box collision attacks. We evaluate CacheAttack in security-critical tasks and agentic workflows. It achieves a hit rate of 86% in LLM response hijacking and can induce malicious behaviors in LLM agent, while preserving strong transferability across different embedding models. A case study on a financial agent further illustrates the real-world impact of these vulnerabilities. Finally, we discuss mitigation strategies.

Multimodal large language models (MLLMs) are increasingly used to automate chart generation from data tables, improving efficiency but introducing new misuse risks. We present ChartAttack, a framework for evaluating how MLLMs use design misleaders to generate charts that induce incorrect interpretations. We also introduce AttackViz, a chart question-answering (QA) dataset labeled with effective misleaders and their induced incorrect answers. ChartAttack reduces MLLM QA accuracy by 17.2 points in-domain and 11.9 points cross-domain. Conditional deception rates show targeted effects: correct answers shift to attacker-intended answers 11.2% of the time in-domain and 11.7-14.9% cross-domain, while originally incorrect answers rarely change. A controlled human study shows that ChartAttack-generated charts also reduce human QA performance. Finally, fine-tuning on AttackViz improves in-domain MLLM robustness to misleading charts. Our findings highlight the need for secure, robust MLLM chart generation. Code and data are publicly available on the project website.

Vulnerability detection methods based on deep learning (DL) have shown strong performance on benchmark datasets, yet their real-world effectiveness remains underexplored. Recent work suggests that both graph neural network (GNN)-based and transformer-based models, including large language models (LLMs), yield promising results when evaluated on curated benchmark datasets. These datasets are typically characterized by consistent data distributions and heuristic or partially noisy labels. In this study, we systematically evaluate two representative DL models-ReVeal and LineVul-across four representative datasets: Juliet, Devign, BigVul, and ICVul. Each model is trained independently on each respective dataset, and their code representations are analyzed using t-SNE to uncover vulnerability related patterns. To assess realistic applicability, we deploy these models along with four pretrained LLMs, Claude 3.5 Sonnet, GPT-o3-mini, GPT-4o, and GPT-5 on a curated dataset, VentiVul, comprising 20 recently (May 2025) fixed vulnerabilities from the Linux kernel. Our experiments reveal that current models struggle to distinguish vulnerable from non-vulnerable code in representation space and generalize poorly across datasets with differing distributions. When evaluated on VentiVul, our newly constructed time-wise out-of-distribution dataset, performance drops sharply, with most models failing to detect vulnerabilities reliably. These results expose a persistent gap between academic benchmarks and real-world deployment, emphasizing the value of our deployment-oriented evaluation framework and the need for more robust code representations and higher-quality datasets.

Large Language Models (LLMs) are increasingly used as interfaces to information, code, and real-world services, making prompt-level security failures a practical concern. Although jailbreak attacks, defenses, datasets, and automated judgers have advanced rapidly, evaluation remains fragmented across threat models, access assumptions, cost budgets, datasets, and success criteria. This makes reported attack success rates and defense gains hard to compare. This SoK systematizes LLM prompt security across concepts, data, tooling, and measurement. We propose linked taxonomies for jailbreak attacks, defenses, and model vulnerabilities, while separating technical mechanisms from attacker and defender capabilities. We also formalize threat, access, and cost assumptions as explicit evaluation metadata. To support reproducible evaluation, we release JailbreakDB, PromptSecurity-Eval, and PromptSecurity, a modular platform that represents each experiment as a tuple of model, attack, defense, dataset, and judger. Using matched evaluations across models, attacks, defenses, and judgers, we show that access regime, native harmful-query behavior, attack cost, defense backfire, taxonomy subcategory, and judger choice all materially affect security conclusions. Together, these artifacts support reproducible, cost-aware, and taxonomy-grounded evaluation of LLM prompt security. Leaderboard: https://datasec-lab.github.io/PromptSecurityLeaderboard/. Dataset: https://huggingface.co/datasets/youbin2014/JailbreakDB. GitHub: https://github.com/datasec-lab/PromptSecurity.

We present a lightweight, model-agnostic temporal layer for RAG and use cybersecurity data to separate two problems that are usually conflated. For freshness, a half-life recency prior surfaces the newest relevant item where a cosine-only baseline scores 0.00; on a hard NVD CVE test, where the freshest item is not the most similar, it reaches Latest@10 of 0.60 versus 0.20 for a semantic-then-newest baseline, but stays partial and parameter-sensitive. For topic evolution, a heuristic tracker's low 0.08 macro-F1 is driven by the labeling rule, not the clusterer (HDBSCAN: 0.10; fixing the rule alone reaches 0.49, and 0.96 without clustering noise). We contribute a reproducible decoupling of the two, with honest real-data scope and a reference implementation.

What fundamentally distinguishes an adversarial attack from a misclassification due to limited model expressivity or finite data? In this work, we investigate this question in the setting of high-dimensional binary classification, where statistical effects due to limited data availability play a central role. We introduce a new error metric that precisely capture this distinction, quantifying model vulnerability to consistent adversarial attacks -- perturbations that preserve the ground-truth labels. Our main technical contribution is an exact and rigorous asymptotic characterization of these metrics in both well-specified models and latent space models, revealing different vulnerability patterns compared to standard robust error measures. The theoretical results demonstrate that as models become more overparameterized, their vulnerability to label-preserving perturbations grows, offering theoretical insight into the mechanisms underlying model sensitivity to adversarial attacks.

Supervised-learning-based vulnerability detectors often fall short due to limited labelled training data. In contrast, Large Language Models (LLMs) are trained on vast unlabelled code corpora, yet perform only marginally better than coin flips when directly prompted to detect vulnerabilities. In this paper, we reframe vulnerability detection as anomaly detection, based on the premise that vulnerable code is rare and thus anomalous relative to patterns learned by LLMs. We introduce ANVIL, which performs a masked code reconstruction task: The LLM reconstructs a masked line of code, and deviations from the original are scored as anomalies. We propose a hybrid anomaly score that combines exact match, cross-entropy loss, prediction confidence, and structural complexity. We evaluate our approach across multiple LLM families, scoring methods, and context sizes, and against vulnerabilities after the LLM's training cut-off. On the PrimeVul dataset, ANVIL outperforms state-of-the-art supervised detectors - LineVul, LineVD, and LLMAO - achieving up to 2x higher Top-3 accuracy, 75% better Normalized MFR, and a significant improvement on ROC-AUC. Finally, by integrating ANVIL with fuzzers, we uncover two previously unknown vulnerabilities, demonstrating the practical utility of anomaly-guided detection.

Large language models (LLMs) are increasingly used for program synthesis, yet they often generate code that is functionally plausible but insecure. Progress in secure code generation has been hindered by benchmarks that are small, non-executable, leak mitigation details, or rely on noisy analyzers and subjective judgments, making it difficult to measure whether security improves without sacrificing correctness. We address these gaps with CodeSecEval, an execution-based benchmark for secure code generation, comprising 255 Python tasks spanning 77 CWE categories. Each task provides paired insecure and secure implementations together with executable functional and vulnerability-targeted security tests, enabling precise and reproducible evaluation of secure code generation and insecure-code repair. Building on CodeSecEval, we propose SecAwareCoder, an agent-based framework that shifts code generation toward secure-by-construction synthesis. SecAwareCoder performs task-adaptive threat modeling to identify security-sensitive regions and derive task-grounded vulnerability hypotheses, uses these hypotheses to guide both constraint-aware code generation and security-aware test synthesis, and leverages execution feedback for targeted refinement. Experiments across multiple LLM backbones show that SecAwareCoder consistently improves Pass@1 and security robustness over prompting and analyzer-driven baselines, narrowing the security--correctness gap in LLM code generation.

Reliability of AI systems is a fundamental concern for the successful deployment and widespread adoption of AI technologies. Unfortunately, the escalating complexity and heterogeneity of AI hardware systems make them increasingly susceptible to hardware faults, e.g., silent data corruptions (SDC), that can potentially corrupt model parameters. When this occurs during AI inference/servicing, it can potentially lead to incorrect or degraded model output for users, ultimately affecting the quality and reliability of AI services. In light of the escalating threat, it is crucial to address key questions: How vulnerable are AI models to parameter corruptions, and how do different components (such as modules, layers) of the models exhibit varying vulnerabilities to parameter corruptions? To systematically address this question, we propose a novel quantitative metric, Parameter Vulnerability Factor (PVF), inspired by architectural vulnerability factor (AVF) in computer architecture community, aiming to standardize the quantification of AI model vulnerability against parameter corruptions. We define a model parameter's PVF as the probability that a corruption in that particular model parameter will result in an incorrect output. In this paper, we present several use cases on applying PVF to three types of tasks/models during inference -- recommendation (DLRM), vision classification (CNN), and text classification (BERT), while presenting an in-depth vulnerability analysis on DLRM. PVF has been a critical metric used for making key error management design decisions in productionizing Meta's in-house AI chip - MTIA.

The growing complexity of software calls for innovative solutions that support the deployment of reliable and secure software. Machine Learning (ML) has shown its applicability to various complex problems and is frequently used in the dependability domain, both for supporting systems design and verification activities. However, using ML is complex and highly dependent on the problem in hand, increasing the probability of mistakes that compromise the results. In this paper, we introduce Propheticus, a ML framework that can be used to create predictive models for reliable and secure software systems. Propheticus attempts to abstract the complexity of ML whilst being easy to use and accommodating the needs of the users. To demonstrate its use, we present two case studies (vulnerability prediction and online failure prediction) that show how it can considerably ease and expedite a thorough ML workflow.

Quantized Vision-Language-Action (VLA) models expose a weight-fault surface: Rowhammer-style faults can corrupt deployed INT8 bits. We present the first bit-flip attack on a VLA: a few gradient-selected flips reduce closed-loop success to $0\%$, while hundreds of random flips are harmless. Across four model variants spanning three action-head families, damaging bits concentrate in a few action-generating layers, but the empirical budget depends sharply on the head: direct regression and token policies fall in $1$--$5$ flips, whereas the evaluated flow-matching policies require ${\sim}100$--$300$. Our fixed-direction manifold-escape loss cuts \pizero{}'s budget from ${\sim}1000$ to ${\sim}100$ flips, and a matched five-direction sweep shows that the attack is not specific to an all-positive direction. On a direct head, protecting $3.1\%$ of weights preserves $60\%$ success at $K{=}100$, and protecting $5.3\%$ moves the open-loop break threshold from 3 to 100 flips. Finally, task-calibrated emulated $K{=}100$ flips yield $0/20$ real-robot successes, versus $14/20$ clean and $16/20$ global-random. Weight integrity is therefore a security boundary for embodied foundation models. Code is included as ancillary material.