Higher Attack Success Rates

Automated red-teaming for LLMs often discovers narrow attack slices, missing diverse real-world threats, and yielding insufficient data for safety fine-tuning. We introduce Persona-Conditioned Adversarial Prompting (PCAP), which conditions adversarial search on diverse attacker personas (e.g., doctors, students, malicious actors) and strategy sets to explore realistic attack scenarios. By running parallel persona-conditioned searches, PCAP discovers transferable jailbreaks across different contexts and generates rich defense datasets with automatic metadata tracking. On GPT-OSS 120B, PCAP increases attack success from 57% to 97% while producing 2-6$\times$ more diverse prompts covering varied real-world scenarios. Critically, fine-tuning lightweight adapters on PCAP-generated data significantly improves model robustness (recall: 0.36 $\rightarrow$ 0.99, F1: 0.53 $\rightarrow$ 0.96) with minimal false positives, demonstrating a practical closed-loop approach from vulnerability discovery to automated alignment.

Diffusion models are the leading approach for tabular data synthesis and are increasingly used to share sensitive records. Whether they actually protect privacy has become a pressing question. Membership inference attacks are the standard tool for this purpose, yet existing attacks assume a single-table setting and ignore the multi-relational structure of real sensitive data. A core challenge in assessing privacy risks from membership inference attacks in multi-table settings is how to leverage auxiliary information from relations associated with the target table, such as its parent tables. Particularly, we study a practical setting in which such auxiliary information is available only when training the attack model. At inference time, the attacker observes only the attribute values of the target record from the target table. We propose FERMI (FEature-mapping for Relational Membership Inference), which resolves this gap by enriching single-table features with relational membership signal. Across three tabular diffusion architectures and three real-world relational datasets, FERMI consistently improves attack performance over single-table baselines, with TPR@$0.1$FPR rising by up to 53% over the single-table baseline in the white-box setting and 22% in the black-box setting.

Decentralized multi-agent path finding (MAPF) routes a team of agents on a shared grid, each acting from its own local view. The standard solution trains one shared neural policy with Proximal Policy Optimization (PPO), a popular on-policy reinforcement learning algorithm. Such a policy works well on clean observations, but a small input perturbation on one agent often changes its action, which then blocks a neighbour, and the team jams. In this paper we present two training recipes that keep the same network and the same deployment loop, yet make the policy hold up under perturbed observations. The first recipe, Adv-PPO, trains the shared policy against worst-case perturbations of its own input and selects the checkpoint by performance under adversarial perturbation. The second recipe, Adv-PPO+MACER, fine-tunes that checkpoint with a small on-policy smoothness term whose gradient follows the certified radius of randomized smoothing. On POGEMA with 8x8 maps and four agents, the unprotected PPO policy reaches 95.8% clean success but only 2.5% under the strongest attack. Adv-PPO recovers worst-case success to 59.2% at one percentage point of clean cost. Adv-PPO+MACER recovers it to 77.5% +/- 6.0% across three independent seeds at less than one percentage point of clean cost. We support these numbers with per-attack curves, a certified action-stability sanity check (which measures the smoothed-policy wrapper, not the deployed argmax policy), and side-by-side rollout storyboards that show the failure mode and the fix inside one environment instance.

AI pentesting agents are increasingly credible as offensive security systems, but current benchmarks still provide limited guidance on which will perform best in real-world targets. Existing evaluation protocols assess and optimize for predefined goals such as capture-the-flag, remote code execution, exploit reproduction, or trajectory similarity, in simplified or narrow settings. These tools are valuable for measuring bounded capabilities, yet they do not adequately capture the complexity, open-ended exploration, and strategic decision-making required in realistic pentesting. In this paper, we present a practical evaluation protocol that shifts assessment from task completion to validated vulnerability discovery, allowing evaluation in sufficiently complex targets spanning multiple attack surfaces and vulnerability classes. The protocol combines structured ground-truth with LLM-based semantic matching to identify vulnerabilities, bipartite resolution to score findings under realistic ambiguity, continuous ground-truth maintenance, repeated and cumulative evaluation of stochastic agents, efficiency metrics, and reduced-suite selection for sustainable experimentation. This protocol extends the state of the art by enabling a more realistic, operationally informative comparison of AI pentesting agents. To enable reproducibility, we also release expert-annotated ground truth and code for the proposed evaluation protocol: https://github.com/ethiack/ethibench.

An attacker can split a malicious goal into sub-prompts that each look benign on their own and only become harmful in combination. Existing LLM safety benchmarks evaluate prompts one at a time, or across turns of a single chat, and so do not look for a malicious signal spread across separate sessions with no shared context. We build FragBench, a benchmark drawn from 24 real-world cyber-incident campaigns, which keeps the full attack trail: the multi-fragment kill chain, the per-fragment safety-judge verdicts, sandboxed execution traces, and a matched set of benign cover sessions. FragBench splits this trail into two paired tasks: an adversarial rewriter that hardens fragments against a single-turn safety judge (FragBench Attack), and a graph-based user-level detector trained on the resulting interactions (FragBench Defense). The single-turn judge is near chance on the released corpus by construction, but four GNN variants and three classical-ML baselines all recover the cross-session feature, reaching aggregate event-level F1 = 0.88-0.96. Defending against fragmented LLM misuse therefore requires modeling the cross-session interaction graph, rather than isolated prompts. Our generator, rewriter, sandbox harness, and detector are released at https://github.com/LidaSafety/fragbench.

Jailbreak attacks -- adversarial prompts that bypass LLM alignment through purely linguistic manipulation -- pose a growing operational security threat, yet the field lacks large-scale, reproducible infrastructure for generating, categorizing, and evaluating them systematically. This paper addresses that gap with three contributions. (1) Large-scale compositional jailbreak dataset. We construct 114,000 adversarial prompts by applying 912 composing strategies to 125 harmful seed prompts from JailBreakV-28K. Every prompt is assigned to one of 14 cybersecurity attack categories (e.g., malware, phishing, privilege escalation) via a six-model majority-vote pipeline, and each strategy is ranked by effectiveness per category, enabling principled strategy selection grounded in concrete adversarial objectives. (2) Automated jailbreak generation. We instruction-fine-tune category-aware LLMs on Moderate and Optimal subsets, producing models that synthesize fluent jailbreak prompts from a harmful seed at inference time -- no templates, no gradient search. Our generators achieve perplexity 24-39 versus 40-140 for AutoDAN and AmpleGCG, with safety-filter evasion rates of 0.29-0.51 Mal (LlamaPromptGuard-2-86M), enabling controllable, scalable red-teaming under realistic adversarial conditions. (3) OPTIMUS: a training-free jailbreak evaluator. OPTIMUS is a continuous metric J(S,H) that jointly captures semantic similarity between the harmful seed and the jailbreak (S) and harmfulness probability (H) via calibrated penalty functions. Unlike binary attack success rate (ASR), OPTIMUS requires no task-specific training, generalizes across evolving strategies, and exposes a stealth-optimal regime (S*=0.57, H*=0.43) that ASR misses. Experiments across 114,000 prompts confirm that OPTIMUS separates Weak, Moderate, and Optimal jailbreaks with category-level evidence binary evaluation cannot supply.

Digital identity verification systems used in remote onboarding rely on document images to authenticate users, making them vulnerable to localized manipulations of key identity fields such as facial photographs and textual information. Existing forgery detection methods, developed primarily for natural-image forensics, show limited transferability to structured identity documents. We propose FLiD, a lightweight field-localized framework that targets critical identity regions rather than processing full-document images. A fine-tuned object detector first localizes face and text fields; a frozen MobileNetV3-Small backbone then extracts compact field-level embeddings, which are classified by lightweight neural network with only 191K trainable parameters. FLiD achieves AUC scores of 0.880 (face), 0.954 (text), and 0.923 (both-field attacks), with corresponding EERs of 18.05%, 11.61%, and 15.16%, representing absolute reductions of 29-35 percentage points over a full-document baseline trained from scratch. FLiD also consistently outperforms general-purpose manipulation detectors (TruFor, MMFusion, UniVAD) across all attack scenarios while requiring 13x fewer parameters and 21x fewer FLOPs

Many jailbreak attack research papers report attack success rates for a limited number of parameter settings, even though there are many combinations of parameter settings that could be used. Further, when new jailbreak papers are released, they often benchmark results against single configurations of existing attacks. This position paper argues such practices are fundamentally insufficient for characterising the threat posed by parameterised jailbreak attacks, and comparing attacks. Most jailbreak attacks expose multiple internal parameters, system prompt templates, conversation rounds, cipher dispersion, teaching shots, and ASR varies substantially across these parameters. Reporting only the best-case configuration discards two pieces of information that defenders genuinely need: how typical that performance is across the variant space, and how much of the attack surface is missed by selecting a single variant. We propose two new measures for jailbreak attacks: the Variant Sensitivity Measure (VSM) and Union Coverage (UC). VSM quantifies how far the best reported ASR deviates from the mean ASR across the tested variant space, UC is the total fraction of prompts resulting in unsafe responses across all tested configurations. We empirically demonstrate the importance of these measures using two attack families across three open-source target models. For PAIR, the best template reaches 69% ASR on Mistral-7B and 75% on Qwen3-0.6B, while UC rises to 88% and 93%, respectively. For bijection on Mistral-7B, the best variant reaches 81% ASR, but the 36-variant union covers 100% of HarmBench-100 prompts. We argue that distributional reporting, publishing VSM alongside ASR and enumerating variant coverage as fully as compute allows, should become the new minimum standard for parameterised jailbreak evaluation.

Large Language Models (LLMs) increasingly employ alignment techniques to prevent harmful outputs. Despite these safeguards, attackers can circumvent them by crafting adversarial prompts. Predominant token-level optimization methods primarily rely on optimizing for exact affirmative templates (e.g., ``\textit{Sure, here is...}''). However, these paradigms frequently encounter bottlenecks such as suboptimal convergence, compromised prompt naturalness, and poor cross-model generalization. To address these limitations, we propose Semantic Representation Attack (SRA), a novel LLM-agnostic paradigm that fundamentally reconceptualizes adversarial objectives from exact textual targeting to malicious semantic representations. Theoretically, we establish the semantic Coherence-Convergence Relationship and derive a Cross-Model Semantic Generalization bound, proving that maintaining semantic coherence guarantees both white-box semantic convergence and black-box transferability. Technically, we operationalize this framework via the Semantic Representation Heuristic Search (SRHS) algorithm, which preserves interpretability and structural coherence of the adversarial prompts during incremental discrete token chunk expansion. Extensive evaluations demonstrate that our framework achieves a 99.71% average attack success rate across 26 open-source LLMs, with strong transferability and stealth.

Modern machine learning deployments increasingly compose specialized models into dynamic inference pipelines, where upstream components produce intermediate predictions that determine the workload and inputs of downstream components. The cost of processing an input is therefore not determined by any single model, but by two coupled factors: the per-inference cost of each invoked component and its workload volume. Because these pipelines run under hard real-time constraints, efficiency is a fundamental requirement for system availability. We show that this structure creates an efficiency-attack surface that existing methods targeting single models cannot exploit: on identical inputs and budgets, path-aware targeting inflates FLOPs by $2,407\times$ while the strongest single-model baseline achieves $117\times$ -- a $20\times$ gap attributable entirely to where the attack is directed. We formalize this as the adversarial path-selection problem and present AESOP, a framework combining vulnerability-guided path ranking with adaptive loss weighting. We evaluate AESOP on five pipelines plus a production-realistic deployment variant with batching, bounded buffering, and confidence-threshold defenses. AESOP achieves up to $2,407\times$ FLOPs and $419\times$ latency inflation in white-box setting and 58$\times$ FLOPs / 17$\times$ latency in gray-box settings. Under system-level defenses, the attack is not neutralized but redirected: pipelines are forced to choose between throughput collapse ($0.578 \to 0.006$ input/s) and $96.7\%$ data loss to sustain throughput.

Tool-calling text-to-image (T2I) agents can plan and execute multi-step tool chains to accomplish complex generation and editing queries. However, this capability introduces a new safety attack surface: harmful outputs may arise from tool orchestration, where individually benign steps combine into unsafe results, making prompt-only jailbreak techniques insufficient. We present OrchJail, an orchestration-guided fuzzing framework for jailbreaking tool-calling T2I agents. Its core idea is to exploit high-risk tool-orchestration patterns: by learning from successful jailbreak tool-calling traces and their causal relationships to prompt wording, OrchJail directly guides the fuzzing search toward prompts that are more likely to trigger unsafe multi-step tool behaviors, rather than relying on surface-level textual perturbations. Extensive experiments demonstrate that OrchJail improves jailbreak effectiveness and efficiency across representative toolcalling T2I agents, achieving higher attack success rates, better image fidelity, and lower query costs, while remaining robust against common jailbreak defenses. Our work highlights tool orchestration as a critical, previously unexplored attack surface and provides a novel framework for uncovering safety risks in T2I agents.

GNNs have become a standard tool for learning on relational data, yet they remain highly vulnerable to backdoor attacks. Prior defenses often depend on inspecting specific subgraph patterns or node features, and thus can be circumvented by adaptive attackers. We propose PRAETORIAN, a new defense that targets intrinsic requirements of effective GNN backdoors rather than surface-level cues. Our key observation is that flipping a victim node's prediction requires substantial influence on the victim: attackers tend to either inject many trigger nodes or rely on a small set of highly influential ones. Building on this observation, PRAETORIAN (i) analyzes internal correlations within potential trigger subgraphs to detect abnormally large injected structures, and (ii) quantifies external node influence to identify triggers with disproportionate impact. Across our evaluations, PRAETORIAN reduces the average attack success rate (ASR) to 0.55% with only a 0.62% drop in clean accuracy (CA), whereas state-of-the-art defenses still yield an average ASR of >20% and a CA drop of >3% under the same conditions. Moreover, PRAETORIAN remains effective against a range of adaptive attacks, forcing adversaries to either inject many trigger nodes to achieve high ASR (>80%), which incurs a >10% CA drop, or preserve CA at the cost of limiting ASR to 18.1%. Overall, PRAETORIAN constrains attackers to an unfavorable trade-off between efficacy and detectability.

The rapid expansion of the Internet of Things (IoT) and Industrial IoT (IIoT) has created a massive, heterogeneous attack surface that challenges traditional network security mechanisms. While Federated Learning (FL) offers a privacy-preserving alternative to centralized Intrusion Detection Systems (IDS), standard approaches struggle to generalize across diverse device behaviors and typically fail to utilize the vast amounts of unlabeled data present in realistic edge environments. To bridge these gaps, we propose CLAD, a holistic framework that seamlessly incorporates Clustered Federated Learning (CFL) with a novel Dual-Mode Micro-Architecture ($\text{DM}^2\text{A}$). This unified approach simultaneously tackles the two primary bottlenecks of IoT security: device heterogeneity and label scarcity. The $\text{DM}^2\text{A}$ component features a shared encoder followed by two branches, enabling joint unsupervised anomaly detection and supervised attack classification; this allows the framework to harvest intelligence from both labeled and unlabeled clients. Concurrently, the clustering component dynamically groups devices with congruent traffic patterns, preventing global model divergence. By carefully combining these elements, CLAD ensures that no data is discarded and distinct operational patterns are preserved. Extensive evaluations demonstrate that this integrated approach significantly outperforms state-of-the-art baselines, achieving a 30% relative improvement in detection performance in scenarios with 80% unlabeled clients, with only half the communication cost.

Today, advances in medical technology extensively utilize 3D volume data for accurate and efficient diagnostics. However, sharing these data across networks in telemedicine poses significant security risks of data tampering and unauthorized copying. To address these challenges, this paper proposes a novel reversible-zero watermarking approach, termed Vol-Mark, for medical volume data to protect their ownership and authenticity in telemedicine. The proposed Vol-Mark method offers two key benefits: 1) it designs a volume data feature extractor that leverages contrastive learning to efficiently extract discriminative and stable volumetric features, ensuring robustness against 3D attacks; 2) it introduces the cubic difference expansion (c-DE) technique, which leverages the 3D integer wavelet transform to embed watermark bits into neighboring voxels within cubes at low-frequency coefficients. The voxel differences within each cube are expanded to create embedding space, and a majority voting mechanism is employed during extraction to enhance reliability. The embedding process incurs low distortion and supports lossless removal, thereby preserving the integrity and diagnostic accuracy of medical volume data. Through these two benefits, Vol-Mark enables both integrity verification and ownership verification. Integrity verification is first performed, and ownership verification through hypothesis testing is further conducted to enhance reliability, particularly under data tampering or watermark removal attacks. Comprehensive experimental results show the effectiveness of the proposed method and its superior robustness against conventional, geometric, and hybrid attacks on medical volume data. In particular, through multiple tasks evaluations, Vol-Mark consistently achieves an ACC above 0.90 in most attack scenarios, outperforming existing methods by a clear margin.

Jailbreak attacks on audio language models (ALMs) optimize audio perturbations to elicit unsafe generations, and they typically update the entire waveform densely throughout optimization. In this work, we investigate the necessity of such dense optimization by analyzing the structure of token-aligned gradients in ALMs. We find that gradient energy is highly non-uniform across audio tokens, indicating that only a small subset of token-aligned audio regions dominates the optimization signal. Motivated by this observation, we propose Token-Aware Gradient Optimization (TAGO), which enables sparse jailbreak optimization by retaining only waveform gradients aligned with audio tokens that have high gradient energy, while masking the remaining gradients at each iteration. Across three ALMs, TAGO outperforms baselines, and substantial sparsification preserves strong attack success rates (e.g. on Qwen3-Omni, $\mathrm{ASR}_{l}$ remains at 86% with a token retention ratio of 0.25, compared to 87% with full token retention). These results demonstrate that dense waveform updates are largely redundant, and we advocate that future audio jailbreak and safety alignment research should further leverage this heterogeneous token-level gradient structure.

Frontier AI safety claims - published assertions that a highly capable general-purpose model is below a threshold of concern, adequately mitigated, or suitable for release - increasingly shape model deployment, governance, and public trust. Yet the artefacts needed to evaluate them are routinely withheld, producing an evidential inversion: the most consequential claims in AI safety are often the least reproducible. This position paper argues that NeurIPS should require reproducibility standards for papers making such claims, treating non-reproducibility not as a transparency preference but as an evaluation-methodology failure. The 2026 International AI Safety Report [Bengio et al., 2026] concludes that reliable pre-deployment safety testing has become harder to conduct and that models now distinguish test from deployment contexts; the 2025 Foundation Model Transparency Index [Wan et al., 2025] reports a sector-average transparency score of 40/100 with no major developer adequately disclosing train-test overlap; contemporaneous measurement-theory work shows that attack-success-rate comparisons across systems are often founded on low-validity measurements [Chouldechova et al., 2025]. We propose a three-tier disclosure framework, distinguishing public, controlled, and claim-restricted disclosure, paired with a mandatory claim inventory, scope statements, and a phased implementation path with graduated sanctions. The framework treats secrecy and openness as endpoints of a spectrum, with controlled review (via a federated colloquium of qualified secure-review hosts) covering claims whose artefacts cannot be released publicly, and right-scaling claims whose artefacts cannot be reviewed even confidentially. The standard the community applies to its most consequential claims should be at least as high as the standard it applies to its least.

Coding agents often pass per-prompt safety review yet ship exploitable code when their tasks are decomposed into routine engineering tickets. The challenge is structural: existing safety alignment evaluates overt requests in isolation, leaving models blind to malicious end-states that emerge from sequenced compliance with innocuous-looking requests. We introduce MOSAIC-Bench (Malicious Objectives Sequenced As Innocuous Compliance), a benchmark of 199 three-stage attack chains paired with deterministic exploit oracles on deployed software substrates (10 web-application substrates, 31 CWE classes, 5 programming languages) that treats both exploit ground truth and downstream reviewer protocol as first-class evaluation axes. On this benchmark, nine production coding agents from Anthropic, OpenAI, Google, Moonshot, Zhipu, and Minimax compose innocuous tickets at 53-86% end-to-end ASR with only two refusals across all staged runs. In a matched direct-prompt experiment over four frontier Claude/Codex agents, vulnerable-output rates fall to 0-20.4%: Claude primarily refuses, while Codex primarily hardens rather than emitting the vulnerable implementation - ticket staging silences both defense modes simultaneously. Downstream, code reviewer agents approve 25.8% of these confirmed-vulnerable cumulative diffs as routine PRs, and a full-context implementation protocol closes only 50% of the staged/direct gap, ruling out context fragmentation as the sole explanation. As a deployable but non-adaptive mitigation, reframing the reviewer as an adversarial pentester reduces evasion across the evaluated reviewer subset; pentester framed evasion ranges from 3.0% to 17.6%, and an open-weight Gemma-4-E4B-it reviewer under this framing detects 88.4% of attacks on the dataset with a 4.6% false-positive rate measured on 608 real-world GitHub PRs.

We show that remotely hosted applications employing in-context learning when augmented with a retrieval function to select in-context examples can be vulnerable to membership-inference attacks even when the service provider and users are separate parties. We propose two black-box membership inference attacks that exploit query text prefixes to distinguish member from non-member inputs. The first attack uses a reference model to estimate an otherwise unavailable loss metric. The second attack improves upon it by eliminating the reference model and instead computing a membership statistic through a simple but novel weighted-averaging scheme. Our comprehensive empirical evaluations consider a stricter case in which the adversary has a paraphrased version of the text in the queries and show that our attacks can exhibit stronger resilience to paraphrasing and outperform three prior attacks in many cases with small number of prefixes. We also adapt an existing ensemble prompting defense to our setting, demonstrating that it substantially mitigates the privacy leakage caused by our second attack.

Large language models (LLMs) employ safety mechanisms to prevent harmful outputs, yet these defenses primarily rely on semantic pattern matching. We show that encoding harmful prompts as coherent mathematical problems -- using formalisms such as set theory, formal logic, and quantum mechanics -- bypasses these filters at high rates, achieving 46%--56% average attack success across eight target models and two established benchmarks. Crucially, the effectiveness depends not on mathematical notation itself, but on whether a helper LLM deeply reformulates the harmful content into a genuine mathematical problem: rule-based encodings that apply mathematical formatting without such reformulation perform no better than unencoded baselines. We introduce a novel Formal Logic encoding that achieves attack success comparable to Set Theory, demonstrating that this vulnerability generalizes across mathematical formalisms. Additional experiments with repeat post-processing confirm that these attacks are robust to simple prompt augmentation. Notably, newer models (GPT-5, GPT-5-Mini) show substantially greater robustness than older models, though they remain vulnerable. Our findings highlight fundamental gaps in current safety frameworks and motivate defenses that reason about mathematical structure rather than surface-level semantics.

Regulatory frameworks such as GDPR increasingly require that ML predictions be accompanied by post-hoc explanations, even when raw data and trained models cannot be released. Differential privacy (DP) is the standard mitigation for the residual privacy risk of releasing these explanations. We show that DP is not sufficient: an adversary observing only DP-perturbed GNN explanations can reconstruct hidden graph structure with high accuracy. Our attack, PRIVX, exploits the fact that the Gaussian DP mechanism is a single DDPM forward step at known noise level σ(ε), recasting reconstruction as reverse diffusion conditioned on the corrupted signal, a principled Bayesian denoiser under known DP corruption. We formalise a stratified adversary model parameterised by (M, \hatε, \hatδ, S, ρ) that interpolates between oblivious and oracle attackers, and derive endpoint-matched two-sided bounds on reconstruction AUC. For practitioners, we provide regime-stratified guidance on explainer choice: on homophilic graphs, neighbourhood-aggregating explainers (GraphLIME, GNNExplainer) leak more structure than per-node gradient explainers under the same DP budget; on strongly heterophilic graphs the ordering reverses. We introduce PRIVF as an auxiliary diagnostic sharing the same diffusion backbone to decompose leakage into explainer-induced and intrinsic graph-distribution components. Experiments across seven benchmarks, three DP mechanisms, and three GNN backbones show PRIVX achieves AUC above 0.7 at ε = 5 on five of seven datasets, with the attack succeeding well within typically deployed privacy budgets.

In Dung-style abstract argumentation, various semantics capture notions of acceptability of arguments. The admissibility semantics capture the notion that an argument can be consistently defended from any potential counterargument. Weak semantics often relax the demands of admissibility by restricting which counterarguments must be taken seriously (e.g., discounting self-defeating or otherwise incoherent attacks). Many prominent proposals for weak semantics remain extension-based in a stronger sense. While these semantics discount attacks from arguments which are considered unreasonable, they still require a uniform defense against all reasonable arguments, even if they are collectively inconsistent. This uniformity can be too demanding when defensibility is inherently strategic, and thus the appropriate reply depends on the opponent's line of attack. We introduce tenability, a family of dialogue-based semantics that formalize when a designated argument (or a set of arguments) can be maintained in debate by a proponent against any conflict-free attack which the opponent may present. The approach is motivated by three natural benchmark patterns: self-defeating attack, floating assignment, and disjunctive reinstatement, on which tenability behaves differently from all weak semantics previously considered in the literature. We define three variants -- static tenability, tenability, and strong tenability -- via monotone commitment games over finite conflict-free moves, differing in the obligations imposed on the disputants. We establish the relative strength of these notions, prove implications and separations with previously studied weak semantics, and we analyze computational complexity on finite frameworks: deciding static tenability is $Π^P_2$-complete, while deciding tenability and strong tenability is PSPACE-complete.

Memory systems enable otherwise-stateless LLM agents to persist user information across sessions, but also introduce a new attack surface. We characterize the Trojan Hippo attack, a class of persistent memory attacks that operates in a more realistic threat model than prior memory poisoning work: the attacker plants a dormant payload into an agent's long-term memory via a single untrusted tool call (e.g., a crafted email), which activates only when the user later discusses sensitive topics such as finance, health, or identity, and exfiltrates high-value personal data to the attacker. While anecdotal demonstrations of such attacks have appeared against deployed systems, no prior work systematically evaluates them across heterogeneous memory architectures and defenses. We introduce a dynamic evaluation framework comprising two components: (1) an OpenEvolve-based adaptive red-teaming benchmark that stress-tests defenses and memory backends against continuously refined attacks, and (2) the first capability-aware security/utility analysis for persistent memory systems, enabling principled reasoning about defense deployment across different usage profiles. Instantiated on an email assistant across four memory backends (explicit tool memory, agentic memory, RAG, and sliding-window context), Trojan Hippo achieves up to 85-100% ASR against current frontier models from OpenAI and Google, with planted memories successfully activating even after 100 benign sessions. We evaluate four memory-system defenses inspired by basic security principles, finding they substantially reduce attack success rates (to as low as 0-5%), though at utility costs that vary widely with task requirements. Because of this substantial security-utility tradeoff, the effective real-world deployment of defenses remains an open challenge, which our evaluation framework is specifically designed to address.

We present MultiBreak, a scalable and diverse multi-turn jailbreak benchmark to evaluate large language model (LLM) safety. Multi-turn jailbreaks mimic natural conversational settings, making them easier to bypass safety-aligned LLM than single-turn jailbreaks. Existing multi-turn benchmarks are limited in size or rely heavily on templates, which restrict their diversity. To address this gap, we unify a wide range of harmful jailbreak intents, and introduce an active learning pipeline for expanding high-quality multi-turn adversarial prompts, where a generator is iteratively fine-tuned to produce stronger attack candidates, guided by uncertainty-based refinement. Our MultiBreak includes 10,389 multi-turn adversarial prompts, spans 2,665 distinct harmful intents, and covers the most diverse set of topics to date. Empirical evaluation shows that our benchmark achieves up to a 54.0 and 34.6 higher attack success rate (ASR)} than the second-best dataset on DeepSeek-R1-7B and GPT-4.1-mini, respectively. More importantly, safety evaluations suggest that diverse attack categories uncover fine-grained LLM vulnerabilities}, and categories that appear benign under single-turn can exhibit substantially higher adversarial effectiveness in multi-turn scenarios. These findings highlight persistent vulnerabilities of LLMs under realistic adversarial settings and establish MultiBreak as a scalable resource for advancing LLM safety.

The emergence of deep neural networks capable of revealing high-fidelity scene details from sparse 3D point clouds has raised significant privacy concerns in visual localization involving private maps. Lifting map points to randomly oriented 3D lines is a well-known approach for obstructing undesired recovery of the scene images, but these lines are vulnerable to a density-based attack that can recover the point cloud geometry by observing the neighborhood statistics of lines. With the aim of nullifying this attack, we present a new privacy-preserving scene representation called \emph{sphere cloud}, which is constructed by lifting all points to 3D lines crossing the centroid of the map, resembling points on the unit sphere. Since lines are most dense at the map centroid, the sphere cloud mislead the density-based attack algorithm to incorrectly yield points at the centroid, effectively neutralizing the attack. Nevertheless, this advantage comes at the cost of i) a new type of attack that may directly recover images from this cloud representation and ii) unresolved translation scale for camera pose estimation. To address these issues, we introduce a simple yet effective cloud construction strategy to thwart new attack and propose an efficient localization framework to guide the translation scale by utilizing absolute depth maps acquired from on-device time-of-flight (ToF) sensors. Experimental results on public RGB-D datasets demonstrate sphere cloud achieves competitive privacy-preserving ability and localization runtime while not excessively compensating the pose estimation accuracy compared to other depth-guided localization methods.

Dynamic quantization emerged as a practical approach to increase the utilization and efficiency of the machine learning serving flow. Unlike static quantization, which applies quantization offline, dynamic quantization operates on tensors at run-time, adapting its parameters to the actual input data. Today's mainstream machine learning frameworks, including ML compilers and inference engines, frequently recommend dynamic quantization as an initial step for optimizing model serving. This is because dynamic quantization can significantly reduce memory usage and computational load, leading to faster token generation and improved model serving efficiency without substantial loss in model accuracy. In this paper, we reveal a critical vulnerability in dynamic quantization: an adversary can exploit such quantization strategy to steal sensitive user data placed in the same batch as the adversary's input. Our analysis demonstrates that dynamic quantization, when improperly implemented or configured, can create side channels that expose information about other inputs within the same batch. We call this phenomenon Quantamination, describing contamination from quantization. Specifically, we show that at least 4 of the most popular ML frameworks in use today either default to or can use configurations that leak data across the batch boundary. This data leakage, in theory, allows attackers to partially or even fully recover other users' batched input data, representing a serious privacy risk for existing ML serving frameworks.

The prevalence of biometric authentication has been on the rise due to its ease of use and elimination of weak passwords. To date, most biometric authentication systems have been designed for on-device authentication of the device owner (e.g., smartphones and laptops). Recently, biometric authentication systems have started to emerge that are designed to authenticate users against cloud databases storing representations of biometrics for large numbers of users (potentially millions), such as those facilitating biometric payments. However, the use of a large cloud database introduces a significant attack vector, as a breach of the database could lead to the compromise of all enrolled users' sensitive biometric data. Indeed, all such existing systems either do not adequately protect against such a breach, or are impractical to deploy and use due to their high computational overhead. In this work, we present a new biometric authentication system that provides provable security guarantees against data breaches, while remaining scalable and performant. To do so, we marry artificial intelligence with advanced cryptographic techniques in a novel fashion, providing several optimizations along the way. Our work is the first to show that real-world scalable privacy-preserving biometric authentication without auxiliary identifiers is feasible, and we believe that it will spur widespread industrial adoption and further research in this area.

The deployment of Large Language Models (LLMs) as assistants in electric grid operations promises to streamline compliance and decision-making but exposes new vulnerabilities to prompt-based adversarial attacks. This paper evaluates the risk of jailbreaking LLMs, i.e., circumventing safety alignments to produce outputs violating regulatory standards, assuming threats from authorized users, such as operators, who craft malicious prompts to elicit non-compliant guidance. Three state-of-the-art LLMs (OpenAI's GPT-4o mini, Google's Gemini 2.0 Flash-Lite, and Anthropic's Claude 3.5 Haiku) were tested against Baseline, BitBypass, and DeepInception jailbreaking methods across scenarios derived from nine NERC Reliability Standards (EOP, TOP, and CIP). In the initial broad experiment, the overall Attack Success Rate (ASR) was 33.1%, with DeepInception proving most effective at 63.17% ASR. Claude 3.5 Haiku exhibited complete resistance (0% ASR), while Gemini 2.0 Flash-Lite was most vulnerable (55.04% ASR) and GPT-4o mini moderately susceptible (44.34% ASR). A follow-up experiment refining malicious wording in Baseline and BitBypass attacks yielded a 30.6% ASR, confirming that subtle prompt adjustments can enhance simpler methods' efficacy.

Agentic AI systems introduce a security surface that is qualitatively different from that of stateless LLMs. They persist memory, invoke external tools, coordinate with peer agents, and operate across sessions, allowing attacks to emerge not only at the prompt interface but also through architectural state, delegated authority, and long-horizon interactions. Existing security taxonomies, however, primarily organize threats by attack type, such as prompt injection or jailbreaking, and therefore obscure where in the agentic stack a threat arises and over what timescale it manifests. We propose the Layered Attack Surface Model (\lasm), a structural taxonomy for agentic AI security. \lasm decomposes the agentic stack into seven layers -- Foundation, Cognitive, Memory, Tool Execution, Multi-Agent Coordination, Ecosystem, and Governance -- and augments them with a four-class temporality axis covering instantaneous, session-persistent, cross-session cumulative, and sub-session-stack threats. We use this 7$\times$4 framework to analyze 116 papers from 2021--2026. The resulting map reveals that the upper layers of the agentic stack remain sharply under-explored, especially for long-horizon and stack-propagating threats; multiple documented attack regions have no corresponding defenses; and current benchmarks provide no coverage for cross-session or sub-session-stack failure modes. We further derive a cross-layer defense taxonomy, defense recipes for canonical attack classes, and a dependency DAG that separates near-term engineering gaps from fundamental research challenges. We release the per-paper coding, robustness scripts, and a reference Agent Bill of Materials schema to support reproducible analysis.

Safety-oriented instruction-following is supposed to keep LLM-controlled robots safe. We show it also creates an availability attack surface. By injecting short safety-plausible phrases (1-5 tokens) into a robots audio channel, an adversary can trigger the models safety reasoning to halt or disrupt execution without jailbreaking the model or overriding its policy. In the embodied setting, this is a semantic denial-of-service attack: the agent stops because the injected signal looks like a legitimate alert. Across four vision-language models, seven prompt-level defenses, three deployment modes, and single- and multi-injection settings, we find that prompt-only defenses trade off attack suppression against genuine hazard response. The strongest defenses reduce hard-stop attack success on some models, but defenses change the form of disruption, not its fact: suppressed hard stops re-emerge as acknowledge loops and false alerts, which we measure with Disruption Success Rate (DSR). We further find that injection variety is consistently more effective than repeating the same phrase, suggesting that models treat diverse safety cues as corroborating evidence. The practical implication is architectural rather than prompt-level: systems that route unauthenticated audio text directly into the LLM create an avoidable security dependency between safety monitoring and action selection.

Semantic Communication (SC) backdoor attacks aim to utilize triggers to manipulate the system into producing predetermined outputs via backdoored shared knowledge. Current SC backdoors adopt monomorphic paradigms with single attack target, which suffers from limited attack diversity, efficiency, and flexibility in heterogeneous downstream scenarios. To overcome the limitations, we propose SemBugger, a polymorphic SC backdoor. By dynamically adjusting the trigger intensity, SemBugger finely-grained controls over the SC knowledge to generate diverse malicious results from the system. Specifically, SemBugger is realized through a multi-effect poisoning-training framework. It introduces graded-intensity triggers to poison training data and optimizes SC systems with hierarchical malicious loss. The trained system's knowledge dynamically adapts to trigger intensity in inputs to yield target outputs, all while preserving transmission fidelity for benign samples. Moreover, to augment SC security, we propose a provable robustness defense that resists SemBugger's homogeneous attacks through a controlled noise mechanism. It operates via strategically adding noise in SC inputs, and we formally provide a theoretical lower bound on the defense efficacy. Experiments across diverse SC models and benchmark datasets indicate that SemBugger attains high attack efficacy while maintaining the regular functionality of SC systems. Meanwhile, the designed defense effectively neutralizes SemBugger attacks.

Deep learning drives major advances in autonomous driving (AD), where object detectors are central to perception. However, adversarial attacks pose significant threats to the reliability and safety of these systems, with physical adversarial patches representing a particularly potent form of attack. Physical adversarial patch attacks pose severe risks but are usually crafted for a single model, yielding poor transferability to unseen detectors. We propose AdvAD, a transfer-based physical attack against object detection in autonomous driving. Instead of targeting a specific detector, AdvAD optimizes adversarial patches over multiple detection models in a unified framework, encouraging the learned perturbations to capture shared vulnerabilities across architectures. The optimization process adaptively balances model contributions and enforces robustness to physical variations. It further employs data augmentation and geometric transformations to maintain patch effectiveness under diverse physical conditions. Experiments in both digital and real-world settings show that AdvAD consistently outperforms state-of-the-art (SOTA) attacks in performance and transferability.

Recommender Systems~(RS) have been shown to be vulnerable to injective attacks, where attackers inject limited fake user profiles to promote the exposure of target items to real users for unethical gains (e.g., economic or political advantages). Since attackers typically lack knowledge of the victim model deployed in the target RS, existing methods resort to using a fixed surrogate model to mimic the potential victim model. Despite considerable progress, we argue that the assumption that \textit{poisoned data generated for the surrogate model can be used to attack other victim models} is wishful. When there are significant structural discrepancies between the surrogate and victim models, the attack transferability inevitably suffers. Intuitively, if we can identify the worst-case victim model and iteratively optimize the poisoning effect specifically against it, then the generated poisoned data would be better transferred to other victim models. However, exactly identifying the worst-case victim model during the attack process is challenging due to the large space of victim models. To this end, in this work, we propose a novel attack method called Sharpness-Aware Poisoning (\textit{SharpAP}). Specifically, it employs the sharpness-aware minimization principle to seek the approximately worst-case victim model and optimizes the poisoned data specifically for this worst-case model. The poisoning attack with SharpAP is formulated as a min-max-min tri-level optimization problem. By integrating SharpAP into the iterative process for attacks, our method can generate more robust poisoned data which is less sensitive to the shift of model structure, mitigating the overfitting to the surrogate model. Comprehensive experimental comparisons on three real-world datasets demonstrate that \name~can significantly enhance the attack transferability.

Aligned large language models (LLMs) remain vulnerable to adversarial manipulation, and their reliance on web-scale pretraining creates a subtle but consequential attack surface. We study Stealth Pretraining Seeding (SPS), a threat model in which adversaries distribute small amounts of poisoned content across stealth websites, increasing the likelihood that such material is absorbed into future training corpora derived from sources such as Common Crawl. Because each individual payload is tiny, diffuse, and superficially benign, the attack is difficult to detect during dataset construction or filtering. The result is a latent form of poisoning that remains largely invisible under standard evaluation, yet can later be activated by a precise trigger such as <00TRIGGER00>. We call this attack PermaFrost, reflecting its latent and reactivatable nature. We study it through PermaFrost-Attack, a controlled framework for latent conceptual poisoning, together with three geometric diagnostics: Thermodynamic Length, Spectral Curvature, and the Infection Traceback Graph. Across multiple model families and scales, we show that this controlled SPS proxy can induce persistent unsafe behavior that often remains hidden under standard evaluation. Our results identify SPS as a practical and underappreciated threat to future foundation models. This paper introduces a novel geometric diagnostic lens for systematically examining latent model behavior, providing a principled foundation for detecting, characterizing, and understanding vulnerabilities that may remain invisible under standard evaluation.

Advancing face morphing attack techniques is crucial to anticipate evolving threats and develop robust defensive mechanisms for identity verification systems. This work introduces DCMorph, a dual-stream diffusion-based morphing framework that simultaneously operates at both identity conditioning and latent space levels. Unlike image-level methods suffering from blending artifacts or GAN-based approaches with limited reconstruction fidelity, DCMorph leverages identity-conditioned latent diffusion models through two mechanisms: (1) decoupled cross-attention interpolation that injects identity-specific features from both source faces into the denoising process, enabling explicit dual-identity conditioning absent in existing diffusion-based methods, and (2) DDIM inversion with spherical interpolation between inverted latent representations from both source faces, providing geometrically consistent initial latent representation that preserves structural attributes. Vulnerability analyses across four state-of-the-art face recognition systems demonstrate that DCMorph achieves the highest attack success rates compared to existing methods at both operational thresholds, while remaining challenging to detect by current morphing attack detection solutions.

Deep learning has emerged as a powerful approach for malware detection, demonstrating impressive accuracy across various data representations. However, these models face critical limitations in real-world, non-stationary environments where both malware characteristics and detection systems continuously evolve. Our research investigates a fundamental security question: Can an attacker generate adversarial malware samples that simultaneously evade classification and remain inconspicuous to drift monitoring mechanisms? We propose a novel approach that generates targeted adversarial examples in the classifier's standardized feature space, augmented with sophisticated similarity regularizers. By carefully constraining perturbations to maintain distributional similarity with clean malware, we create an optimization objective that balances targeted misclassification with drift signal minimization. We quantify the effectiveness of this approach by comprehensively comparing classifier output probabilities using multiple drift metrics. Our experiments demonstrate that similarity constraints can reduce output drift signals, with $\ell_2$ regularization showing the most promising results. We observe that perturbation budget significantly influences the evasion-detectability trade-off, with increased budget leading to higher attack success rates and more substantial drift indicators.

Current text anonymization evaluation relies on span-based metrics that fail to capture what an adversary could actually infer, and assumes a single data subject, ignoring multi-subject scenarios. To address these limitations, we present SPIA (Subject-level PII Inference Assessment), the first benchmark that shifts the unit of evaluation from text spans to individuals, comprising 675 documents across legal and online domains with novel subject-level protection metrics. Extensive experiments show that even when over 90% of PII spans are masked, subject-level inference protection drops as low as 33%, leaving the majority of personal information recoverable through contextual inference. Furthermore, target-subject-focused anonymization leaves non-target subjects substantially more exposed than the target subject. We show that subject-level inference-based evaluation is essential for ensuring safe text anonymization in real-world settings.

Security Operations Centers (SOCs) increasingly encounter difficulties in correlating heterogeneous alerts, interpreting multi-stage attack progressions, and selecting safe and effective response actions. This study introduces AgentSOC, a multi-layered agentic AI framework that enhances SOC automation by integrating perception, anticipatory reasoning, and risk-based action planning. The proposed architecture consolidates several layers of abstraction to provide a single operational loop to support normalizing alerts, enriching context, generating hypotheses, validating structural feasibility, and executing policy-compliant responses. Conceptually evaluated within a large enterprise environment, AgentSOC improves triage consistency, anticipates attackers' intentions, and provides recommended containment options that are both operationally feasible and well-balanced between security efficacy and operational impact. The results suggest that hybrid agentic reasoning has the potential to serve as a foundation for developing adaptive, safer SOC automation in large enterprises. Additionally, a minimal Proof-Of-Concept (POC) demonstration using LANL authentication data demonstrated the feasibility of the proposed architecture.

As LLMs continue to shape real-world applications, automated jailbreak generation becomes essential to reveal safety weaknesses and guide model improvement. Existing automatic jailbreak generation methods have not yet fully considered two important aspects: adaptability to evolving safety-finetuned models, which affects their effectiveness on newer model versions, and diversity in generated prompts, which can cause narrow or repetitive attack patterns. To address these issues, we propose EvoJail, an instruction-fusion-driven evolutionary jailbreak generation framework that formalizes jailbreak prompt generation as a multi-objective black-box optimization problem and leverages the principles of evolutionary algorithms to search for jailbreak prompts that can adapt across different model versions and exhibit diverse attack patterns. Specifically, EvoJail integrates jailbreak prompt generation into an iterative evolutionary loop, where at each iteration candidate prompts are evaluated directly against the target model and then selected and varied based on the target model's responses, enabling the generation process to continuously adapt to model updates. To enhance diversity, EvoJail introduces field-aware instruction fusion to construct diverse starting points and incorporates diversity-aware objectives into the evolutionary fitness function, guiding the search toward prompts with richer semantic variation, while further designing multi-level LLM-based mutation operators that modify prompt structures at different granularities to promote structural diversity throughout the evolutionary process. Results demonstrate that EvoJail has stronger adaptability and can achieve over $93\%$ attack success rate and more than $5.6\%$ improvement in diversity metrics over state-of-the-art methods.

Tool-integrated agents are deployed on the premise that external tools ground their outputs in reality. Yet this very reliance creates a critical attack surface. Current evaluations benchmark capability in benign settings, asking "can the agent use tools correctly" but never "what if the tools lie". We identify this Trust Gap: agents are evaluated for performance, not for skepticism. We formalize this vulnerability as Adversarial Environmental Injection (AEI), a threat model where adversaries compromise tool outputs to deceive agents. AEI constitutes environmental deception: constructing a "fake world" of poisoned search results and fabricated reference networks around unsuspecting agents. We operationalize this via POTEMKIN, a Model Context Protocol (MCP)-compatible harness for plug-and-play robustness testing. We identify two orthogonal attack surfaces: The Illusion (breadth attacks) poison retrieval to induce epistemic drift toward false beliefs, while The Maze (depth attacks) exploit structural traps to cause policy collapse into infinite loops. Across 11,000+ runs on five frontier agents, we find a stark robustness gap: resistance to one attack often increases vulnerability to the other, demonstrating that epistemic and navigational robustness are distinct capabilities.

Today, machine learning is widely applied in sensitive, security-related, and financially lucrative applications. Model extraction attacks undermine current business models where a model owner sells model access, e.g., via MLaaS APIs. Additionally, stolen models can enable powerful white-box attacks, facilitating privacy attacks on sensitive training data, and model evasion. In this paper, we focus on Decision Trees (DT), which are widely deployed in practice. Existing black-box extraction attacks for DTs are either query-intensive, make strong assumptions about the DT structure, or rely on rich API information. To limit attacks to the black-box setting, CPU vendors introduced Trusted Execution Environments (TEE) that use hardware-mechanisms to isolate workloads from external parties, e.g., MLaaS providers. We introduce TrEEStealer, a high-fidelity extraction attack for stealing TEE-protected DTs. TrEEStealer exploits TEE-specific side-channels to steal DTs efficiently and without strong assumptions about the API output or DT structure. The extraction efficacy stems from a novel algorithm that maximizes the information derived from each query by coupling Control-Flow Information (CFI) with passive information tracking. We use two primitives to acquire CFI: for AMD SEV, we follow previous work using the SEV-Step framework and performance counters. For Intel SGX, we reproduce prior findings on current Xeon 6 CPUs and construct a new primitive to efficiently extract the branch history of inference runs through the Branch-History-Register. We found corresponding vulnerabilities in three popular libraries: OpenCV, mlpack, and emlearn. We show that TrEEStealer achieves superior efficiency and extraction fidelity compared to prior attacks. Our work establishes a new state-of-the-art for DT extraction and confirms that TEEs fail to protect against control-flow leakage.

Vision-Language Models (VLMs) are increasingly used in clinical diagnostics, yet their robustness to adversarial attacks remains largely unexplored, posing serious risks. Existing medical attacks focus on secondary objectives such as model stealing or adversarial fine-tuning, while transferable attacks from natural images introduce visible distortions that clinicians can easily detect. To address this, we propose MedFocusLeak, a highly transferable black-box multimodal attack that induces incorrect yet clinically plausible diagnoses while keeping perturbations imperceptible. The method injects coordinated perturbations into non-diagnostic background regions and employs an attention distraction mechanism to shift the model's focus away from pathological areas. Extensive evaluations across six medical imaging modalities show that MedFocusLeak achieves state-of-the-art performance, generating misleading yet realistic diagnostic outputs across diverse VLMs. We further introduce a unified evaluation framework with novel metrics that jointly capture attack success and image fidelity, revealing a critical weakness in the reasoning capabilities of modern clinical VLMs.

Instructional alignment, the match between intended cognition and enacted activity, is central to effective instruction but hard to operationalize at scale. We examine alignment in cybersecurity simulations using multimodal traces from 23 teams (76 students) across five exercise sessions. Study 1 codes objectives and team emails with Bloom's taxonomy and models the completion of key exercise tasks with generalized linear mixed models. Alignment, defined as the discrepancy between required and enacted Bloom levels, predicts success, whereas the Bloom category alone does not predict success once discrepancy is considered. Study 2 compares predictive feature families using grouped cross-validation and l1-regularized logistic regression. Text embeddings and log features outperform Bloom-only models (AUC0.74 and 0.71 vs. 0.55), and their combination performs best (Test AUC0.80), with Bloom frequencies adding little. Overall, the work offers a measure of alignment for simulations and shows that multimodal traces best forecast performance, while alignment provides interpretable diagnostic insight.

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.

Existing LLM safety evaluations rely on binary attack-success rates and domain-agnostic taxonomies, leaving regulated Banking, Financial Services, and Insurance (BFSI) deployments exposed to failures elicited through legally or professionally plausible framing. We introduce RAHS (Risk-Adjusted Harm Score), a risk-sensitive metric jointly capturing disclosure severity, disclaimer mitigation, and inter-judge agreement, and FinRedTeamBench, a 989-prompt benchmark spanning seven BFSI risk areas and 34 sub-categories mapped to regulatory frameworks. Evaluation uses an ensemble of three heterogeneous LLM judges, validated against human experts, and an adaptive multi-turn red-teaming pipeline. On nine open-weight models, RAHS preserves separation under near-ceiling ASR, ranking is stable under hyperparameter sweeps, and multi-turn pressure drives not only more jailbreaks but more operationally severe disclosures, exposing failure modes that single-turn, domain-agnostic evaluations cannot reveal.

Membership inference attacks (MIAs) threaten the privacy of machine learning models by revealing whether a data point was used during training. Existing MIAs often assume access to public datasets, shadow models, confidence scores or the training distribution, which makes them vulnerable to defenses like confidence masking. Label-only MIAs avoid these assumptions but require thousands of queries per sample. We propose a cost-effective label-only MIA framework based on transferability and model extraction. Querying the target $M$ with active sampling, perturbation-based selection and synthetic data, we extract a surrogate $S$ on which membership inference is performed offline. This shifts query overhead to a one-time extraction phase. It also removes the restriction that defines the label-only setting: the attacker controls $S$ and can read its posteriors and training trajectory, so attacks that cannot be run against $M$ can be run against $S$. On Location, Purchase and Texas, the strongest attack on $S$ improves AUC over the direct attack on $M$ by $0.9$, $5.6$ and $5.0$ percentage points, and improves the true positive rate at $1\%$ false positive rate by $2.8\times$ to $6.3\times$. We characterize how leakage transfer depends on surrogate fidelity, evaluate standard defenses, and report preliminary results on image datasets.

Vision-Language Models (VLMs) are now a core part of modern AI. Recent work proposed several visual jailbreak attacks using single/ holistic images. However, contemporary VLMs demonstrate strong robustness against such attacks due to extensive safety alignment through preference optimization, e.g., reinforcement learning from human feedback (RLHF). In this work, we identify a new vulnerability: while VLM pretraining and instruction tuning generalize well to split-image inputs, safety alignment is typically performed only on holistic images and does not account for harmful semantics distributed across multiple image fragments. Consequently, VLMs often fail to detect and reject harmful split-image inputs, in which unsafe cues emerge only upon combining images. We introduce novel split-image visual jailbreak attacks (\textbf{SIVA}) that exploit this misalignment. Unlike prior optimization-based attacks, which exhibit poor black-box transferability due to architectural and prior mismatches across models, our attacks evolve in progressive phases from naive splitting to an adaptive white-box attack, culminating in a black-box transfer attack. Our strongest strategy leverages a novel adversarial knowledge distillation \textbf{(Adv-KD)} algorithm to substantially improve cross-model transferability. Evaluations on four state-of-the-art modern VLMs and three jailbreak datasets demonstrate that our strongest attack achieves up to 44% higher transfer success than existing baselines. Lastly, we propose efficient ways to address this critical vulnerability in the current VLM safety alignment.

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.

AI agents are autonomous systems that combine LLMs with external tools to solve complex tasks. While such tools extend capability, improper tool permissions introduce security risks such as indirect prompt injection and tool misuse. We characterize these failures as unbalanced tool-driven agency. Agents may retain unnecessary permissions (excessive agency) or fail to invoke required tools (insufficient agency), amplifying the attack surface and reducing performance. We introduce AgenTRIM, a framework for detecting and mitigating tool-driven agency risks without altering an agent's internal reasoning. AgenTRIM addresses these risks through complementary offline and online phases. Offline, AgenTRIM reconstructs and verifies the agent's tool interface from code and execution traces. At runtime, it enforces per-step least-privilege tool access through adaptive filtering and status-aware validation of tool calls. Evaluating on the AgentDojo benchmark, AgenTRIM substantially reduces attack success while maintaining high task performance. Additional experiments show robustness to description-based attacks and effective enforcement of explicit safety policies. Together, these results show that AgenTRIM provides a practical, capability-preserving approach to safer tool use in LLM-based agents.

Input transformation-based attacks improve adversarial transferability by aggregating gradients over transformed inputs. Existing analyses mainly explain their efficacy from image diversity, semantic preservation, attention variance or hypothesis space augmentation, yet overlook the critical role of model frontend responses. In this paper, we revisit transformation-based attacks from an implicit ensemble perspective: each transformation can be viewed as a pre-processing operator before the surrogate model, inducing a distinct frontend response for gradient aggregation. Based on this view, we propose FRO, a Frontend Response-Oriented input transformation method that enriches such responses through two complementary operators. The Local Scaling Operator perturbs local content sampling via block-wise stretch-and-shrink operations, while the Projection Operator modifies global spatial organization through coherent perspective deformation. Together, they produce structured transformed views to optimize transferable adversarial perturbations. Experiments on an ImageNet subset show that FRO consistently improves black-box transferability across diverse CNN and Vision Transformer models. We further analyze the effect of implicit ensemble size and evaluate different transformation-based methods under a unified ensemble scale, demonstrating the superiority of designing input transformations from the perspective of front-end response ensembles.

Existing data poisoning attacks on retrieval-augmented generation (RAG) systems scale poorly because they require costly optimization of poisoned documents for each target phrase. We introduce Eyes-on-Me, a modular attack that decomposes an adversarial document into reusable Attention Attractors and Focus Regions. Attractors are optimized to direct attention to the Focus Region. Attackers can then insert semantic baits for the retriever or malicious instructions for the generator, adapting to new targets at near zero cost. This is achieved by steering a small subset of attention heads that we empirically identify as strongly correlated with attack success. Across 18 end-to-end RAG settings (3 datasets $\times$ 2 retrievers $\times$ 3 generators), Eyes-on-Me raises average attack success rates from 21.9 to 57.8 (+35.9 points, 2.6$\times$ over prior work). A single optimized attractor transfers to unseen black box retrievers and generators without retraining. Our findings establish a scalable paradigm for RAG data poisoning and show that modular, reusable components pose a practical threat to modern AI systems. They also contribute to interpretability research by revealing a strong link between attention concentration and model outputs.

Horizontal federated learning (HFL) backdoor audits often summarize model behavior through clean accuracy (CA), mean attack success rate (ASR), or a single known-trigger test. Such summaries can hide a different failure mode, in which one target label is activated by many trigger realizations. We study this failure mode with FLAT, a latent-conditioned reliability stress test for HFL backdoors. In FLAT, compromised clients still submit ordinary classifier updates to the server, while an attacker-side generator $G(x,t,z)$ separates target intent $t$ from trigger realization $z$. This separation shifts the audit question from whether one known trigger succeeds to how the hidden behavior varies across targets, latent samples, defenses, and post-stop rounds. On CIFAR-10, CIFAR-100, and Tiny-ImageNet, FLAT preserves clean utility while reaching 99.49%, 99.66%, and 94.10% single-target FedAvg ASR. The evaluation also reveals non-uniform defense responses, where a server rule can suppress one target mode while leaving another active. These observations motivate HFL backdoor audits that report target-wise ASR, worst-target ASR, target coverage, latent-sampled behavior, post-stop persistence, and defense response.

Large language models (LLMs) are widely applied in various fields of society due to their powerful reasoning, understanding, and generation capabilities. However, the security issues associated with these models are becoming increasingly severe. Jailbreaking attacks, as an important method for detecting vulnerabilities in LLMs, have been explored by researchers who attempt to induce these models to generate harmful content through various attack methods. Nevertheless, existing jailbreaking methods face numerous limitations, such as excessive query counts, limited coverage of jailbreak modalities, low attack success rates, and simplistic evaluation methods. To overcome these constraints, this paper proposes a multimodal jailbreaking method: JMLLM. This method integrates multiple strategies to perform comprehensive jailbreak attacks across text, visual, and auditory modalities. Additionally, we contribute a new and comprehensive dataset for multimodal jailbreaking research: TriJail, which includes jailbreak prompts for all three modalities. Experiments on the TriJail dataset and the benchmark dataset AdvBench, conducted on 13 popular LLMs, demonstrate advanced attack success rates and significant reduction in time overhead.

The increasing demand for data privacy, alongside the benefits of aggregating data from networked devices, has catalyzed the emergence of federated learning (FL). In FL, clients jointly train a global model by sharing gradients computed over private data. While this paradigm eliminates the need to exchange raw data, inference attacks can still be launched to extract sensitive information from gradients. To this end, partial gradient encryption has emerged as a promising design for balancing privacy and efficiency in practical FL systems, as encrypting only the classification-head gradients is believed to prevent known inference attacks while avoiding the high computational cost of encrypting the entire model. However, this design provides a false sense of privacy. By proposing GDBR, we show that sharing even a single unencrypted layer of gradients can lead to serious privacy leakage. GDBR is the first attack capable of high-fidelity label recovery with partial access to the gradients. It exploits a vulnerability in a commonly used neural building block, constructs a gradient bridge from the unencrypted layer to the final output layer, and approximates the logits information for accurate inference of private labels. These inferred labels not only reveal sensitive information about a client's private dataset but also serve as a prerequisite for many downstream attacks, such as data reconstruction and membership inference. GDBR brings these threats squarely into scope for FL systems employing partial encryption. In addition to theoretical analysis, extensive experiments demonstrate the severity of the problem across a wide variety of datasets and model architectures, including convolutional and transformer-based networks. Overall, our findings challenge the widespread assumption that encrypting only the output layer suffices for privacy protection.

Vision-Language-Action (VLA) models map multimodal perception and language instructions to executable robot actions, making them particularly vulnerable to behavioral backdoor manipulation: a hidden trigger introduced during training can induce unintended physical actions while nominal task performance remains intact. Prior work on VLA backdoors primarily studies untargeted attacks or task-level hijacking, leaving fine-grained control over individual actions largely unexplored. In this work, we present DropVLA, an action-level backdoor attack that forces a reusable action primitive (e.g., open_gripper) to execute at attacker-chosen decision points under a realistic pipeline-black-box setting with limited data-poisoning access, using a window-consistent relabeling scheme for chunked fine-tuning. On OpenVLA-7B evaluated with LIBERO, vision-only poisoning achieves 98.67%-99.83% attack success rate (ASR) with only 0.31% poisoned episodes while preserving 98.50%-99.17% clean-task retention, and successfully triggers the targeted action within 25 control steps at 500 Hz (0.05 s). Text-only triggers are unstable at low poisoning budgets, and combining text with vision provides no consistent ASR improvement over vision-only attacks. The backdoor remains robust to moderate trigger variations and transfers across evaluation suites (96.27%, 99.09%), whereas text-only largely fails (0.72%). We further validate physical-world feasibility on a 7-DoF Franka arm with pi0-fast, demonstrating non-trivial attack efficacy under camera-relative motion that induces image-plane trigger drift. These results reveal that VLA models can be covertly steered at the granularity of safety-critical actions with minimal poisoning and without observable degradation of nominal performance.

Chain-of-thought (CoT) monitors are reported by their aggregate accuracy on a pool of reward hacks. We show that this number is a false average. On Terminal Wrench, about 77% of hacks are given away by the actions alone, and the monitor's pooled accuracy is dominated by them; on the remaining 23%, where the reasoning is the only signal, the same monitor is fragile. We expose the fragility with a controlled attack: we rewrite only the agent's reasoning to read as good-faith engineering, leaving every command and output byte-identical, so the exploit is unchanged. One gradient-free rewrite drops a held-out monitor's catch rate on that subset from about 95% to between 4 and 11%, while the pooled rate falls only about 25 points, the subset's share of the pool times its collapse. The rewrite stays truthful about what happened and lies only about intent, so a monitor that reads only the trace recovers little, even when warned about the attack, and only information from outside the trace helps much. The attack transfers across monitor and agent families, and the dataset's own hack labels show the same gradient with no monitor in the definition. Live agents on 27 tasks produce hollow hacks unaided, but a monitor calibrated on each model's own honest solves catches nearly all of them (AUC at least 0.996 under the narration instruction, 0.97 without it); a small residual has the blindspot's shape, low scores on both actions and narration, though none falls below the calibrated threshold. A linear probe on a surrogate monitor's activations out-reads its verdict on the obfuscated hacks (AUC up to 0.98), but a behavioral control shows this edge does not, by itself, mean the monitor knows more than it says. Monitor accuracy should be reported on the reasoning-dependent subset, where the defense is supportive.