Papers

The NIS-2 Directive mandates robust Risk Management from thousands of small and medium enterprises. To ensure compliance, companies rely on established standards such as the German IT-Grundschutz (IT-GS) of the Federal Office for Information Security. However, IT-GS certification is resource-intensive and requires a high level of manual effort for documentation, validation, and revision, making scalable implementation difficult and expensive. Building upon our previous conceptual framework, this paper presents the technical implementation and empirical evaluation of a Multi-Agent System (MAS) architecture combined with Hybrid Retrieval Augmented Generation (HybridRAG) for the partial automation of IT-GS certification. We introduce two novel technical contributions to the MAS architecture to enforce the compliance rigor. The Hypothesis-Verification Loop in the Structural Analysis (SA) phase that cross-references agent-inferred dependencies against the Knowledge Graph to reduce hallucinations, and a Decoupled Reasoning Pipeline that separates agent-driven semantic extraction from the deterministic protection need inheritance. We utilize the BSI's "RecPlast GmbH" case study as a human expert-generated reference data set for end-to-end evaluation of the architecture and to quantify Precision, Recall, and F1-scores. The performance of the system is investigated across the phases of SA, Protection Needs Assessment (PNA), Modeling, and IT-GS Check. The empirical results reveal noticeable differences throughout the different steps of IT-GS. While the MAS demonstrates high efficacy in semantic tasks (SA and Modeling), significantly reducing manual effort through automated information extraction, quantitative results reveal limitations in logical reasoning phases (PNA and IT-GS Check) as the probabilistic nature of current LLMs struggles to meet the deterministic rigor required by IT-GS.

This paper presents a novel approach to perform semi-automated BSI IT-Grundschutz certification using a MultiLarge Language Model system (MLS) with Hybrid RetrievalAugmented Generation (HybridRAG). Facing the challenges of the Network and Information Security Directive 2 (NIS2) directive, a shortage of specialists, and high implementation costs, our MLS architecture aims to increase efficiency, reduce costs, and support certifiers in maintaining the quality of security concepts while meeting the increased demand for certifications of newly affected companies. The system combines Large Language Models (LLMs) and Knowledge Graphs (KGs) to support different phases of the certification process, including protection needs assessment, modeling, IT-Grundschutz check, measure consolidation, and subsequent realization. Our architecture addresses the growing demand for security concepts and offers an approach to handle the digital security challenges introduced by NIS2.

As graph neural networks (GNNs) become standard tools for critical tasks in circuit design and analysis, their security and privacy risks require careful attention. Here, we present the first comprehensive evaluation of gradient leakage attacks (GLAs) on GNNs in circuit-design and hardware-security tasks, a practical threat that has been largely overlooked. We assess state-of-the-art (SOTA) GNNs, including GraphSAGE, GCN, GIN, and GAT, trained on standard netlist benchmarks (ISCAS'85, EPFL, and TrustHub), for their fundamental vulnerability to GLAs. We find that GLAs can expose sensitive information, such as gate types and distinctive properties of hardware Trojans, which may assist adversaries in analyzing logic locking schemes or evading Trojan detection mechanisms. Our analysis shows that these risks are influenced by architectural features, with attention mechanisms (GAT) exacerbating leakage, while injective aggregation (GIN) provides comparatively stronger resilience. We further evaluate several SOTA defense techniques, including differential privacy, gradient clipping, secure aggregation, model compression with quantization, and adversarial training. We find that these techniques improve resilience only in specific settings and can also compromise model performance. Overall, our work provides key insights toward privacy-preserving GNNs and highlights the need for more robust and efficient defenses. We release our full methodology and artifacts.

Manual formal analysis of cryptographic schemes is labor-intensive and requires substantial expertise. While model-checking tools (e.g., Scyther and Tamarin) and computational-security tools (e.g., CryptoVerif and EasyCrypt) improve the automation of security proofs, they still rely on experts to abstract schemes and write tool-specific formal descriptions. Large language models (LLMs) are a promising alternative, but their effectiveness in this domain remains unexplored due to the absence of standardized evaluation methodologies. To fill this gap, we introduce CrypFormBench (C.F.B for short), a comprehensive benchmark jointly covering symbolic and computational security to evaluate five core LLM capabilities: interpretation, generation, completion, transformation, and correction. It comprises 700 instances spanning 677 schemes, 7 mainstream formal verifier languages, and 160 security properties. The evaluation of 9 state-of-the-art LLMs reveals that most of them perform well on interpretation and completion, given their code-awareness advantages, but struggle with generation, transformation, and correction. Overall, their performance remains limited, with Claude-3.5 achieving the highest score at 48.7 out of 100. We further provide practical guidance, e.g., few-shot prompting, Pass@K sampling, and lightweight fine-tuning, to mitigate the executability bottleneck and improve tool-usable outputs. Taken together, our benchmark and analyses offer a grounded view of current progress and concrete directions toward reliable LLM-assisted formal cryptographic analysis.

Retrieval-Augmented Generation (RAG) has emerged as a dominant paradigm for enhancing large language models with external knowledge. By coupling retrieval mechanisms with generative models, RAG systems improve factual grounding and adaptability across domains. However, integrating retrieval pipelines introduces new security and privacy risks that extend beyond conventional language modeling threats. Sensitive information may be exposed through retrieval indices, query logs, context construction, or federated updates, while adversarial manipulation of knowledge bases can undermine trust in generated outputs. This survey provides a comprehensive examination of privacy and security challenges across RAG systems deployed in centralized, on-device (Micro-RAG), federated, and hybrid paradigms. We present a unified taxonomy of threat surfaces spanning the retrieval, context construction, and generation stages and systematically analyze attack classes, including membership inference, index inference, poisoning, gradient leakage, and collusion. We further review architectural, algorithmic, and cryptographic defenses, highlighting privacy-utility trade-offs and deployment considerations. Finally, we outline open research challenges toward building trustworthy, secure, and resilient RAG systems for real-world applications.

Almost every paper on LLM jailbreaks and prompt injection reports an attack-success rate (ASR), and that number is assigned not by people but by an automated judge: either a safety classifier trained for the task, or a general chat model prompted to grade. The judge is rarely checked. We check it. Using 596 human-labeled completions from the HarmBench classifier validation set, we compare the two judge families against human majority votes and then attack them. The two families fail in opposite ways. The dedicated classifier over-flags (precision 0.835, recall 0.974); three different LLM-as-judges keep high precision (0.81 to 0.94) but show erratic recall (0.06 to 0.65), so the same responses produce very different ASR depending on which judge scores them. The two families also differ sharply in robustness. Wrappers that leave the harmful text untouched and only add benign framing flip every LLM-judge between 57% and 100% of the time, and a single prepended refusal sentence accounts for much of this (39% to 88%). The dedicated classifier resists these surface attacks (at most 6.7%), but a white-box GCG attack on its open weights flips 70% of confident true positives (21 of 30; 95% CI 54 to 86%) even at a small optimization budget. A two-annotator audit confirms the attacks leave the harm intact: every one of 80 sampled flips still contained the harmful content. Because a large and growing share of reported ASR comes from LLM-judges, many such numbers are unreliable both on average and under deliberate pressure. We recommend that papers report judge precision and recall on a human-labeled slice, report ASR corrected for judge precision, and include an adversarial check of the judge. Our code is released.

Persistent memory is a recently proposed memory paradigm that delivers many system-wide benefits, including improved runtime efficiency and the ability of programs to recover from power outages and system crashes. While recent research has investigated techniques for proving functional correctness of programs running on related architectures, this is not the case for the orthogonal concept of information flow security. In this paper, we provide an information flow logic for an unstructured language (i.e., with gotos rather than loops) modelling a simple assembly language. We apply this logic to x86 assembly using a notion of reordering interference freedom (rif) to reason about potential out-of-order propagation of instructions to memory. We then show how this same notion of rif can be used to similarly reason about information flow on persistent memory.

Large Language Models (LLMs) are increasingly used for automated vulnerability detection, but it remains unclear how program structure and semantics should be represented for LLM-based reasoning. Most prompting-based approaches provide raw source code, implicitly assuming that more source-level context gives the model better evidence. This paper challenges that assumption through RepBench, an empirical benchmark comparing raw source code with static-analysis-based program representations. RepBench converts real-world C/C++ vulnerability testcases into multiple representations: raw source, Abstract Syntax Trees (ASTs), Control-Flow Graphs (CFGs), Program Dependence Graphs (PDGs), their combinations, and an auxiliary track of enriched PDGs (ePDGs). Using a curated PrimeVul-derived corpus of 107 Joern-based testcases across five CWE categories, we evaluate ten representation variants under a fixed Chain-of-Thought and structured-output protocol, plus 19 additional ePDG cases generated through VulChecker/Hector. Representation choice substantially affects LLM vulnerability reasoning. The strongest variant, AST+PDG, achieves 83.2% accuracy, compared with 53.5% for raw source. At the family level, graph-only prompts outperform both source-only and source-plus-graph prompts while requiring far less prompt overhead. These results reveal a context dilution effect: adding raw source code to compact structural graph evidence can degrade reasoning by making vulnerability-relevant evidence less salient. Overall, our findings show that carefully selected structural representations offer a better accuracy-overhead tradeoff than simply giving LLMs more raw input, and suggest that static analysis can serve as an effective prompt-construction layer for security-focused LLM reasoning.

In secure two-party Transformer inference, linear layers are typically evaluated using Fully Homomorphic Encryption (FHE) through plaintext-ciphertext or ciphertext-ciphertext matrix multiplications, where key switching primarily occurs and dominates computational overhead in both FHE-based and hybrid FHE-MPC systems. Existing optimizations rely heavily on packing-specific algorithms, limiting their general applicability. Targeting this overhead from a packing-independent perspective, we propose a preprocessing-assisted method for secure attention computation. By decomposing attention into precomputable operations and online interactions, this method reduces online inference-phase key switching without modifying existing packing strategies. However, the first method shifting key switching offline introduces additional storage requirements. To address this, we propose storage-communication trade-off techniques that replace large precomputed ciphertexts with modest online communication, enabling flexible deployment under varying resource constraints. While ciphertext-ciphertext matrix multiplication is offloaded to the preprocessing phase in hybrid schemes and the first layer of FHE-based schemes, these operations still persist in the offline stage and subsequent FHE layers. To further optimize it, we propose a fused key-switch technique targeting the multiplication-followed-by-rotation pattern, which frequently arises in existing RNS-CKKS matrix multiplication schemes. By combining relinearization and rotation into a single procedure, this technique reduces the associated computation costs. Analytical evaluations demonstrate that our proposed techniques significantly reduce online key-switch overhead and provide flexible trade-offs between storage and communication without requiring modifications to existing packing strategies.

Large Language Models (LLMs) have shown promise for automated penetration testing, yet existing end-to-end black-box evaluations are highly susceptible to error cascading: failures in early reconnaissance can mask an agent's actual ability to exploit vulnerabilities. To more accurately characterize these capabilities, we propose a two-stage decoupled evaluation framework that separates exploit execution from reconnaissance. Using ground-truth injection and knowledge-driven ablation across 70 high-fidelity web vulnerability testbeds, our framework isolates exploitation performance from reconnaissance noise. We empirically evaluate five open-source penetration-testing agents, covering multiagent, monolithic, and graph-driven architectures, on a strictly aligned subset of 50 representative vulnerabilities. The results reveal a substantial capability gap. With accurate vulnerability context, agents achieve a functional success rate of up to 90.0%, whereas autonomous reconnaissance, measured by targeted vulnerability recall, plateaus at approximately 50.0%, primarily due to failures in parsing unstructured telemetry. Cross-architectural analysis further reveals distinct capability niches: multi-agent isolation is more effective for long-sequence interactions such as de-serialization, while monolithic and graph-driven designs perform better on short-chain injections and cross-session access-control vulnerabilities, respectively. This decoupled evaluation work provides a fine-grained benchmarking protocol and an empirical basis for designing next-generation automated offensive security agents.

Group signatures are privacy preserving signature schemes in which a group member can anonymously sign messages on behalf of the group, while providing accountability, by allowing the signature of a misbehaving group member be “opened” and the identity of the signer be revealed. In group signature members are admitted to the group by a (trusted) group manager. We motivate the need for a flexible mechanism in applications, such as privacy preserving access in smart environments, and propose a two-level member-join group signature that we call SPonsored Group Signature (SPGS) where group members of level 1 can “sponsor” new members, in level 2, to join the group. This relaxation of user join comes with additional accountability mechanisms: we require that the signature of a sponsored member can be opened to the identity of the sponsor (that is sponsor is responsible for the sponsored member), and while all signatures are anonymous, for the sponsored members, the signatures are linkable. This allows a sponsor to efficiently identify an undesirable sponsored member. We formalize SPGS scheme, define its security using a game-based approach, and give a generic construction of SPGS that uses a (dynamic) group signature scheme, a commitment scheme, and a knowledge-sound non-interactive zero knowledge proof of knowledge, and prove its security. We also give an instantiation of our construction. To show applicability of SPGS in practice, we consider the problem of providing guest access in a smart building, and introduce Anonymous Guest Access Token (AGAT) that allows a temporary guest to anonymously access (a subset of) the building resources. We show how SPGS can be used (together with an IND-CPA secure public key encryption scheme) to give a direct construction for AGAT, and show the efficiency of our guest access protocol when it is instantiated with existing schemes.

As AI systems are rapidly integrated into critical economic, governmental, and national security functions, the gap between AI adoption and AI security readiness continues to widen. This paper presents a prioritized agenda for advancing AI security, informed by structured interviews with leaders across industry, government, and civil society, and refined through a multi-sector expert workshop. Participants identified and ranked the highest-importance and most cost-effective areas where progress could strengthen AI security - from protecting frontier AI systems and their underlying infrastructure to improving cybersecurity practices as AI reshapes the threat landscape. The resulting priorities are organized across four themes: establishing strategic foundations and policy frameworks; advancing public-private coordination and institutional infrastructure; advancing technical security engineering and assurance; and governing agentic AI under adversarial pressure. For each priority area, expert authors provide detailed analyses that define the problem, assess the current landscape, and identify actionable projects that stakeholders across sectors can pursue. The paper aims to serve as an initial practical foundation for coordinated investment and action across the AI security field. It is designed to serve both current practitioners and individuals and organizations looking to enter the field by identifying concrete, high-impact contributions suited to a range of strengths and capacities.

LLMs are increasingly deployed in security-critical systems across healthcare, finance, education, and decision support, yet their inability to forget creates serious cybersecurity, privacy, and safety risks. Sensitive personal information, copyrighted material, hazardous domain knowledge, and memorized training data remain encoded across billions of parameters long after deployment, leaving models vulnerable to extraction, jailbreak attacks, membership inference, and regulatory non-compliance. Real-world incidents, from chatbots regenerating private information to fabricated legal citations producing direct legal and financial cost, place the problem at the center of the emerging-threats landscape rather than the realm of speculation. Because retraining billion-parameter models on revised corpora is computationally infeasible, and because knowledge within an LLM is distributed and entangled across parameters rather than localized to identifiable units, LLM unlearning has emerged as the principal cyber defense response, aiming to remove or suppress targeted knowledge from a trained model without retraining and without eroding what the model should still know. A central question, however, remains unresolved. Do current methods genuinely remove knowledge, or do they only stop the model from expressing it under ordinary prompting conditions? This survey examines LLM unlearning through the lens of security, robustness, and verifiable forgetting, with primary focus on gradient-based methods, which have come to dominate the field due to their compatibility with existing training pipelines and their scalability to billion-parameter models.

Legal requirements might prevent organizations from sharing sensitive data like medical or financial details of consumers which prevents them from leveraging cloud based ML-as-a-service solutions provided by third party providers, which are quickly gaining popularity these days. In this project, we aim to perform inference tasks in Computer Vision in a privacy-preserving manner, i.e, by only looking at encrypted data. Recent advances in fully homomorphic encryption make this possible. A fully homomorphic encryption allows an arbitrary sequence of additive and multiplicative operations to be performed on encrypted data directly. Applying homomorphic encryptions to CNNs requires modifying the conventional CNN layers, so that they adhere to the encryption scheme. Our aim was to explore the best methods to create CNNs which can classify encrypted images directly. We used Microsoft SEAL for performing homomorphic encryption. The performance of these "encryption based CNNs" should be comparable with baseline accuracies of the same CNNs trained on unencrypted data, and the aim was to achieve as low of a hit on inference-time performance as possible. We successfully obtained minimal drop in classification accuracy for various datasets. We used MNIST as our baseline, which is popularly used in related research work and then explored more complex datasets like Kuzushiji MNIST, Fashion-MNIST and CIFAR-10 as a part of our contribution. Additionally, we also added support for more complex operations on top of TenSEAL, like processing colored images (multi-channel input), applying multiple convolutional layers and performing average pooling.

The increasing use of AI systems for code generation raises a central security question: what can today's models and coding agents actually do to produce secure code, where do they still fail, and what would move the field forward? Existing work has explored prompting, fine-tuning, reinforcement learning, and agentic workflows for secure code generation, but the field still lacks a systematic understanding of how these techniques improve security and why substantial failures persist. In this SoK, we systematize the progress, pitfalls, and paths forward for AI secure code generation. We introduce a three-level framework that measures models' natural-language understanding of secure coding principles, their code-level actuation of those principles during generation, and the knowledge–actuation gaps between the two. We instantiate this framework across models and coding agents on benchmarks covering both isolated function-level security and full web-application security. Our results show that secure-coding-principle understanding is a statistically strong predictor of code-level outcomes, including functional correctness, security, and joint functional-security correctness. Yet substantial knowledge–actuation gaps remain: models can recognize relevant security principles but still fail to translate them into secure and functional code. These findings offer a principle-centered account of where AI secure code generation stands today and identify concrete paths forward through principle-guided generation, evaluation, benchmarking, and agentic workflows.

Speculative decoding accelerates inference by letting a draft model propose tokens for a target model to verify, raising a concrete safety question: at temperature zero, can draft-side behavior leak into safety-scored outputs? We answer with Typical-Acceptance Invariance Screen (TAIS), a behavioral-equivalence screen that pairs target-only and speculative outputs on the same safety battery and requires byte-identity evidence, TOST equivalence at +/-3pp, and per-task Cohen's h below a calibrated null cutoff of |h| < 0.1. Applied to a 16,783-sample confirmatory core plus 44,066 matched expansion samples (fp16/bf16 execution, canonical and DPO-adversarial drafts, GPTQ-4bit drafts, two seeds, and four safety benchmarks), the tested temperature-zero vLLM stacks show no detectable safety divergence under TAIS. The largest absolute Cohen's h on matched target-only versus speculative refusal is 0.024, roughly an order of magnitude below the conventional trivial-effect floor; 25 of 27 per-task TOST contrasts pass at the +/-3pp margin (the two non-pass contrasts are capability-domain Wald-CI edge cases at identical ceiling rates, not genuine non-equivalence); the DPO-adversarial draft produces byte-identical output to the canonical draft across 4,006 samples; and bf16 changes 36%-53% of output bytes without moving any per-task safety rate outside equivalence. A separate 4,006-sample 70B production-scale probe, which lacks a matched 70B target-only arm and is therefore not counted as a TAIS pass, produces AdvBench refusal 0.839 over 700 AdvBench completions with 95% Wilson CI [0.809, 0.864]. We make no claim about sampling temperatures, untested frameworks, untested model families, or tree-speculation variants such as EAGLE and Medusa.

Black-box LLMs (accessible only via API) are vulnerable to distillation attacks, in which an attacker queries the model and trains a student on its outputs. A recent line of work proposes output perturbation defenses that modify the teacher's output to reduce student performance while preserving utility for legitimate users. As a relatively new family of approaches, output perturbation defenses lack a shared threat model, making it difficult to compare them, reason about composing them with other attacks, or evaluate their robustness against realistic adversaries. This underspecification matters beyond technical evaluation: when defenses are deployed to protect intellectual property or justify regulatory compliance, an imprecise threat model can create a false sense of security. We propose a threat model framework that describes attackers along three dimensions: a query budget, a data budget, and an interface profile that captures how attackers interact with the API. Using antidistillation sampling as a case study, we show that whether the defense is considered effective depends on the assumed threat model. We argue that future work on distillation defenses, along with any governance or policy frameworks built around them, should explicitly specify and stress-test attacker capabilities along our three dimensions.

In machine learning, model certification has been identified as an important method for gaining assurance about a model's trustworthiness and quality. A model's quality is largely determined by its ability to generalize, i.e., to perform well on data beyond what it was trained on. It is not possible to certify generalization directly, however, as it depends on unknown data and is not directly measurable. Proxies such as test accuracy can be misleading when the training process is perturbed (intentionally or accidentally), and metrics such as sharpness – which has an empirically supported link to generalization – are computationally expensive and can also serve as unreliable signals when training deviates from a prescribed procedure. In this work, we propose directional sharpness, a metric designed to efficiently and reliably indicate generalization despite potential training deviations. We provide empirical and analytical evidence that directional sharpness (1) correlates more strongly with generalization than existing metrics and (2) identifies models with poor generalization more reliably than existing metrics. Furthermore, directional sharpness is efficiently computable in model auditing settings, where the verifier has access to training data, and via zero-knowledge proofs that certify quality without revealing training data.

This paper introduces HelpBench, a benchmark for assessing whether LLMs are capable of providing accurate help in response to questions about digital privacy, safety, and security. We curated 450 questions representing authentic user situations and developed rubrics for each question to evaluate the factual accuracy and tone of a response. Example questions touch on how to regain access to lost or suspended accounts, how to balance the trade-offs of hardware security keys versus other forms of two-factor authentication, whether a suspicious email is likely a scam, or whether an abuser might be able to track an individual based on their device peripherals. We then developed and applied an auto-rater to evaluate responses from 18 state-of-the-art LLMs. Our results indicate that while models provide high-quality advice (with scores of 82% on average), one in ten responses from models scores less than 65%, reflecting inaccurate and even harmful advice. Addressing these failures is critical for models to serve as trustworthy sources of assistance for digital privacy, safety, and security needs.

Malware analysis is a critical aspect of modern cybersecurity. The prevailing industry practice, sandboxing, involves executing suspicious binaries within isolated virtual machines in large-scale data centers. However, this approach can unintentionally expose samples to public platforms such as VirusTotal and MalwareBazaar, and it is both resource-intensive and time-consuming. Burnyard addresses these limitations through a lightweight binary emulation platform that captures observable runtime behavior and records it as structured CSV event traces.

Pseudorandom states, introduced by Ji, Liu, and Song (CRYPTO 2018), are quantum analogues of classical pseudorandom generators. A fundamental property of classical pseudorandom generators is that their output can be stretched to arbitrary polynomial length. Whether an analogous stretching property holds for quantum pseudorandom states has been elusive. Here, we essentially resolve this question for the single-copy variant of pseudorandom states, 1PRS. We show, via a unitary oracle separation, that no fully black-box construction can stretch the output length of a 1PRS by any constant factor greater than one.

Fuzzing is widely used for software and hardware verification, offering an effective alternative to random testing. While gray-box fuzzers benefit from full visibility into the system under test and can leverage execution feedback such as branch coverage, these approaches are not applicable when verifying systems whose firmware or binaries are not publicly available. In such scenarios, obtaining coverage information for guiding the fuzzer becomes infeasible. In this paper, we introduce PowerFuzz, a statistical black-box fuzzing framework that leverages power side-channel measurements as a substitute for binary instrumentation, requiring no internal visibility into the target firmware. A central challenge in black-box firmware fuzzing is determining the executed branches during test execution. To address this challenge, we use power traces to identify branches utilizing a sliding window followed by a growing window full-trace correlation method. This approach also enables the construction of a high-level control-flow graph of the black-box firmware, which we utilize to drive the fuzzer to unexplored execution paths. Extensive evaluation using three embedded hardware platforms and ten firmware benchmarks demonstrates that PowerFuzz can provide branch coverage comparable (within 13.5%) to gray-box fuzzers while significantly outperforming (up to 22%) state-of-the-art black-box fuzzers.

Full-system rehosting plays a critical role in the security analysis of Linux-based firmware. It matches commonly deployed firmware with sufficient background knowledge. However, for custom devices, existing approaches struggle to handle initialization and runtime obstacles in the rehosting process caused by specialized architectures and hardware-dependent configuration, which heavily rely on expert intervention. This ultimately creates fundamental bottlenecks and results in low rehosting efficiency. To address the above challenges, we propose FirmCure, the first LLM-driven full-system rehosting framework designed for autonomous and adaptive rehosting of Linux-based firmware. FirmCure develops an Adaptive Perception Inference mechanism to extract firmware structural dependencies via static analysis, followed by a Reflective Synthesis module for iterative configuration optimization, and finally an Autonomous Runtime Intervention module for real-time error remediation through runtime fault diagnosis and monitoring. We evaluated 21 IoT firmware images from 10 vendors across 5 architectures, while FirmCure achieved a 100% network port opening rate and 90.5% service interactivity, substantially outperforming state-of-the-art baselines. Our experiments confirm that FirmCure's intervention strategies generalize across heterogeneous firmware. The framework successfully reproduces known vulnerabilities and discovers new security flaws.

The use of agentic systems to perform offensive security operations has moved from a theoretical possibility to a commoditized capability. However, while the community has focused on creating more and more capable agents, less attention has been allocated to assessing the security of those systems. In this work, we present the first in-depth security analysis of the most widely used agentic systems for offensive security operations. We show that most of these tools share common design flaws that enable an active adversary to exfiltrate API keys, establish persistent footholds, and fully compromise the operator's machine, even when the agent operates inside a sandboxed container. To support our analysis, we introduce a full cyber kill chain for such agentic systems, capturing the progression from initial LLM manipulation to lateral movement, persistence, guardrail bypass, and sandbox escape. Building on our security analysis, we derive a robust architecture for agentic offensive-security tools and propose actionable, broadly applicable design principles that mitigate the disclosed attack paths at the architectural level.

We show that random linear codes (RLCs) possess nearly optimal discrepancy-type properties in a broad range of settings. Our main results are two general discrepancy theorems: one controls all translates of a fixed test, and the other controls large families of Fourier-pseudorandom tests. Two motivating examples follow: First, RLCs behave essentially like unstructured random codes for list-decoding from errors above capacity. More precisely, an RLC $C\subseteq \mathbb{F}_q^n$ of rate $1 - \frac{1}{n}\log_q|B_\rho| + \varepsilon$, where $|B_\rho|$ is the volume of a radius-$\rho$ Hamming ball in $\mathbb{F}_q^n$, satisfies $|C \cap B| = (1\pm o(1)) \frac{|C|\cdot |B|}{q^n}$ simultaneously for all radius-$\rho$ Hamming balls $B$ with high probability. This vastly generalizes the previously best known fact that RLCs of this rate have covering radius at most $\rho n$ with high probability (Blinovsky, 1987). Second, over prime fields, RLCs behave essentially like unstructured random codes for zero-error list-recovery, and list-recovery from erasures, above capacity. More precisely, for a prime $q>2$ and input list size $2\leq \ell\leq q-1$, an RLC $C\subseteq \mathbb{F}_q^n$ of rate $1-\log_q \ell+\varepsilon$ will satisfy $|C \cap S| = (1\pm o(1)) \frac{|C|\cdot \ell^n}{q^n}$ simultaneously for all combinatorial rectangles $S=S_1\times S_2\times\cdots\times S_n$, where $|S_i|=\ell$ for all $i$, with high probability. An analogous result also holds when we can bound $|S_i|$ only for some of the $i$'s. We use this to show the abundance of locally leakage-resilient $n$-party linear ramp secret sharing schemes with any linear reconstruction threshold and sublinear threshold gap $O(n/\log n)$ over fields of polynomial size $q=\Theta(n^\gamma)$ for a constant $\gamma\in(0,1/5)$. Prior work was stuck at reconstruction thresholds above $n/2$ for both threshold and ramp schemes.

Kubernetes has become the industry standard for orchestrating containers in microservice-based software architectures. While several hardening guidelines and scanning tools for securing Kubernetes clusters and deployments have emerged in recent years, their differing guidance and outputs often lead to inconsistent configuration and prioritization decisions. This work presents a systematic comparison of eight commonly used Kubernetes hardening guidelines. Through this comparison and the inclusion of best practices, we established a benchmark of 79 Kubernetes configuration recommendations and conducted the a structured empirical evaluation of ten popular static configuration scanning tools and their scoring outputs. Our findings reveal substantial disparities in the coverage of configuration issues across hardening guidelines and scanners, as well as inconsistencies in how configuration issues are scored and ranked by different scanners. These results highlight the need for more standardized, transparent, and consistent approaches to risk and severity assessment of Kubernetes configuration issues.

AI security agents increasingly rely on Retrieval-Augmented Generation (RAG) to use external security knowledge for vulnerability analysis and exploit reasoning. This creates a new risk: poisoned write-ups can be operationalized into incorrect exploit behavior. Yet, prior work on RAG poisoning has mostly studied answer corruption in QA settings, much less is known about action-taking security agents. This paper aims to reveal such characteristics with crafted poisons about real-world challenges and AI agents. First, we demonstrate how a crafted single poisoned write-up injected into public-style security knowledge sources which we denote as Poisoned Playbooks, alters the behavior of RAG-based AI security agents. Across 11 CTF challenges, 3 frontier LLM families, 2 model generations, and 11 real-world CVEs, we find that poison adoption is systematic rather than random. To explain this pattern, we introduce the Verification Boundary (VB), a 3-level empirical classification based on what evidence the agent can use to refute a retrieved claim. Finally, we evaluate verification prompting and multi-source retrieval, showing that both help when stronger evidence exists, but weaken under sparse-evidence and zero-day conditions.

Fully Homomorphic Encryption (FHE) enables computations to be performed directly on encrypted data while preserving data confidentiality. However, its practical applications remain limited by high computational costs and development complexity. This paper presents ComputeFHE, an open-source C++ library that facilitates the development of privacy-preserving applications based on the TFHE cryptosystem. The library provides encrypted integer and fixed-point data types together with arithmetic, logical, comparison, conditional, and oblivious array-access operations which allow developers to implement algorithms using a familiar imperative programming paradigm. ComputeFHE supports both conventional TFHE arithmetic based on standard two-input logic gates and an optimized Arithmetic Logic Unit (ALU) architecture utilizing FHE-friendly logic primitives. Experimental results demonstrate significant reductions in the number of required bootstrapping operations, achieving performance improvements of up to 3.9x for selected operations. In addition, the library includes a simulation mode that enables testing, debugging, and complexity analysis without performing actual cryptographic computations while providing circuit complexity and bootstrapping costs. Built on top of OpenFHE, ComputeFHE offers a practical and accessible framework for developing and evaluating privacy-preserving algorithms and applications.

LLM agents increasingly rely on persistent long-term memory, which creates a critical vulnerability that we study here: memory poisoning. An adversary can store untrusted content in one session that later steers a consequential action, such as a payment, a setting change, or data exfiltration, in a future session. Existing defenses base a memory item's authority to act on either its content (detection or trust-scoring) or its derivation history (lineage). We show that both signals are malleable. An attacker can launder an untrusted origin through three channels specific to LLM agents: the agent's own summarization, a trusted-tool echo, and manufactured corroboration. Each makes the content look benign and breaks or flips its derivation edge to “trusted.” We formalize malleability for the memory write-retrieve-act pipeline and prove a machine-checked separation theorem. No content- or lineage-based defense is sound under laundering (T1), write-time origin binding is necessary (T2), and non-malleable origin-bound authority with Sybil-resistant corroboration-gated elevation is sufficient (T3). Our construction, TMA-NM (Tamper-evident Memory Authority, Non-Malleable), instantiates non-malleable information-flow control (IFC) for LLM-agent memory. A cross-defense, cross-attack, and cross-model benchmark over eight frontier models shows that existing defenses fail exactly where the theory predicts (up to 68% laundering attack-success), while TMA-NM reaches 0% attack success on both direct and laundering attacks across all models and channels, at full legitimate utility. We release the benchmark, harness, and machine-checked TLA+ models to support reproducibility.

Large language model (LLM) agents increasingly automate complex tasks by integrating language models with external tools and environments. However, their autonomy poses significant safety risks: agents may execute destructive commands, leak sensitive data, or violate domain constraints. Existing safety approaches face a fundamental tradeoff: hand-crafted rules are interpretable but brittle, with overly conservative rules blocking safe operations (high false positives) while permissive rules miss unsafe behaviors (high false negatives). Neural classifiers lack the interpretability required for safety-critical deployments. We present AutoSpec, a framework that automatically evolves deployed expert-designed safety rules from user safe/unsafe annotations through counterexample-guided inductive synthesis (CEGIS) guided by inductive logic programming (ILP). Starting from the expert rules and a stream of annotated traces, AutoSpec iteratively evaluates rules, mines false-positive and false-negative counterexamples, uses ILP to learn which predicates discriminate them, generates candidate rule edits, and verifies candidates to select the best revision. The key insight is that ILP efficiently identifies predicates that appear frequently in false negatives but rarely in false positives (or vice versa), dramatically pruning the exponential search space of rule edits. This continues until convergence, producing interpretable rules that balance precision and recall. We evaluate AutoSpec on 291 execution traces spanning code execution and embodied agent domains. AutoSpec raises rule F1 to 0.98 and 0.93 across the two domains, achieving up to 94% false positive reduction while maintaining high recall, and converges within 4-5 iterations. The ILP-guided approach achieves up to 4.8x higher F1 than heuristic CEGIS. The learned rules are human-readable, auditable, and generalize to unseen scenarios.