CyberSec Research

Agents using the Model Context Protocol (MCP) rely on semantic matching to select tools from third-party servers, exposing a semantic supply-chain risk through attacker-controlled metadata and outputs. We introduce A2M (Attraction-to-Manipulation), a two-stage black-box framework for hijacking MCP agents. The Attraction phase optimizes tool metadata to increase invocation probability; the Manipulation phase uses execution traces to refine adversarial tool returns that steer agents toward attacker-desired outcomes. On LiveMCPBench, direct attacks optimized and evaluated on GLM-4.6 achieve a macro-average malicious tool invocation rate of 93.6% across four scenarios, increase weighted token costs to 32.4$\times$ the benign baseline under Cognitive Denial of Service, and attain a mean attack success rate of 74.4% across Information Exfiltration, Environment Integrity Compromise, and Reasoning Derailment. Transfer to four other models without re-optimization yields corresponding macro-averages of 63.6%, 2.7$\times$, and 24.5%. These findings motivate stronger tool vetting and runtime isolation in MCP ecosystems. Code is publicly available at https://github.com/Lilaizhen/A2M.

Large language models (LLMs) are increasingly applied to the automated repair of C/C++ security vulnerabilities, and compile rate is a commonly reported proxy for progress: whether the generated patch compiles. We argue that compile rate is a scientifically unreliable metric for single-function vulnerability repair, and we support this with five controlled experiments over 203 vulnerable functions from Big-Vul, three open-source code LLMs (350M to 6.7B parameters), and three prompting strategies. Compile rate (i) barely responds to an intervention that substantially improves the generated code; (ii) is dominated by evaluation-harness and dataset artifacts rather than model quality, with about 64% of compile failures not attributable to the model, a share that is nearly invariant across models; (iii) shifts by 1.8 to 2.7 times on identical patches under a single compiler-standard flag, with zero regressions; (iv) ranks the three models in the opposite order to reference-similarity metrics; and (v) rewards non-repairs when used as an optimization target, since a compiler-feedback loop raises compile rate while similarity to the human fix falls, with manual inspection finding deletion- and placeholder-style non-repairs among the newly compiling outputs. The natural fallback, whole-function CodeBLEU, also fails: an unchanged copy of the vulnerable input outscores every model. We also examine diff_F1, a change-aware screen that scores only the edited region. It gives exactly zero credit to a no-op and near-zero credit to some, though not all, of the deletion-based gaming patches we observed, while still crediting genuine partial edits, so it may serve as a cheap screen before deeper, execution-based analysis. It is not a repair-quality metric, and we report where it falls short. Our findings argue for change-aware, execution-grounded evaluation of LLM-based vulnerability repair.

Generative AI (GenAI) applications have flourished enabling users to chat with large language models, and to create agents to act on their behalf for a variety of tasks. The pace of development of capabilities in this field is incredibly fast with security and safety taking a back seat. Unfortunately, the slower pace at which security and safety mechanisms have evolved has led to real incidents. Policy enables the definition of desirable behavior of applications, and for that reason, it is a cornerstone of making systems secure and compliant. Policy however means different things to different practitioners creating confusion and siloed solutions that are not adequate for compliance. This paper takes a tour of the good, the bad and the ugly when it comes to policy enforcement in GenAI applications. We propose a methodology to systematically analyze and dissect existing approaches to define and enforce policy found in the wild. Based on this principled analysis, we provide recommendations and call for action for the community to address. This paper is a companion extension of USENIX Security 2026 Enigma talk titled "From Alignment to Access Control: A Unified View of GenAI Policy Enforcement" by the author Nathalie Baracaldo.

Even though privacy policies are the primary mechanism organizations use to disclose how they collect, process, and share personal data, they are difficult for average users to interpret, perhaps by design, due to their verbosity and dense legal language. Importantly, there is a lack of standardized metrics that characterize key qualities of a privacy policy beyond regulatory requirements. Recent advances in large language models (LLMs) make it feasible to automatically structure and analyze these documents at scale. In this study, we develop and evaluate an end-to-end, LLM-enabled system that converts raw privacy policies into fine-grained structured representations and a set of quantitative measures. Our pipeline applies a detailed taxonomy to extract specific data elements and governing practices, capturing relational links that connect each practice to the data elements it references. We apply our framework to a diverse corpus of 10,000 website privacy policies, yielding, to the best of our knowledge, the most comprehensive dataset of its kind to date. Building on our structured representations, we introduce the first standardized and repeatable quantitative metrics for evaluating privacy policies along four dimensions: completeness, transparency, commitment to user protection, and emphasis on business-driven data practices. This allows us to compare policies within and across industry sectors, and to assess the tension between user protection and business interests.

The rapid capability gains of frontier language models are widely attributed to improved reasoning abilities, yet this cannot be verified as raw CoT traces in closed-source systems are hidden. By registering a simple custom tool through a standard API feature, we induce frontier models to externalize intermediate reasoning. Because these traces may reflect post-hoc rationalization rather than genuine reasoning, we first evaluate against native CoT on open-source models and extend to closed-source frontier models including GPT-6 Astra. We find that the extracted reasoning matches native reasoning performance and substantially outperforms no-reasoning baselines, across competition mathematics, science, and code generation. We then characterize how frontier models structure their intermediate reasoning. Across token efficiency, reasoning-step types, and induced reasoning trees, we identify systematic differences in how models externalize, compress, and organize reasoning. We find that Astra exhibits token-efficient directed reasoning, selecting a correct trajectory earlier, while resolving elementary steps internally and externalizing only crucial reasoning. These findings provide a behavioral lens on frontier-model reasoning beyond benchmark scores.

Honeypots are designed to deceive attackers, and recent work shows they can also derail autonomous LLM-based pentesters. These evaluations, however, largely consider attackers unaware of the deception they face. We study the opposite setting: an autonomous attacker explicitly equipped to recognize and act on honeypot fingerprints. We introduce Rouxii, an AI-driven penetration-testing framework that integrates counter-deception into reconnaissance and pivots from honeypot detection to exploitation. We evaluate matched vanilla and anti-deception Rouxii configurations across three reasoning models and eleven network setups over twelve cycles (1,544 attack reports). Between the matched cohorts, which differ only in the prompt, counter-deception raises correct honeypot identification from 19% to 97%, an effect strongest on OT services (11% to 97%), while false alarms on the real service stay at 0.7%. Deception-unaware baselines (PentestGPT, HackingBuddy) fail similarly, indicating the effect is not specific to our framework. Detection, moreover, is not the endpoint: through a white-box analysis of the honeypots themselves we show that a detected trap can be turned against its operator, demonstrating a denial-of-service that disables Conpot without tripping its liveness monitoring, and a corruption of the intelligence a GasPot instance reports. These findings show that deception effectiveness depends strongly on attacker knowledge, and that evaluations of honeypot resilience against AI attackers must account for adversaries that actively reason about and exploit the deception layer.

Secure multiparty computation (MPC) enables mutually distrustful parties to compute on private digital inputs. We initiate the study of spatiotemporal MPC, extending this paradigm to functionalities whose inputs additionally depend on physical facts such as the parties' locations, times, or trajectories. Such protocols must simultaneously hide spatiotemporal information and ensure its physical consistency: a malicious party should not be able to make the functionality operate on a spatiotemporal input inconsistent with its actual physical state. The main conceptual challenge is to formulate extraction of spatiotemporal information within the simulation-based security framework. We introduce arguments of spatiotemporal knowledge, following the principle "I prove, therefore I am:" rather than defining physical presence directly through a mathematical relation, we define it operationally through the ability to complete a sound spatiotemporal verification protocol. Accordingly, an extractor recovers a spatiotemporal point from a successful prover and certifies its physical validity by using the extracted prover to succeed in an auxiliary spatiotemporal verification protocol. Building on this notion, we define universally composable (UC) security for spatiotemporal MPC, capturing privacy, physical consistency, and composability. We provide constructions achieving this new MPC notion. We first construct UC-secure commit-and-prove of spatiotemporal knowledge: in the CRS model under LWE against quantum provers without pre-shared entanglement, and in the QROM against quantum provers with unbounded pre-shared entanglement. Using these protocols, we obtain UC-secure spatiotemporal MPC from semi-honest post-quantum MPC. We also extend our framework to support quantum functionalities with classical spatiotemporal input.

The Terrapin attack against SSH channel integrity (USENIX Security 2024) used a novel attack vector: attacks on the channel state. Surprisingly, not all AEAD modes of SSH were equally affected by this attack, and it remained an open question if "unaffected" meant "secure". Existing formal models for secure channels are based on stateful encryption. However, these models do not define what the channel state is and how it is used as input to the different AEAD modes. In this paper, we propose a formal model for channel integrity under partially chosen state. Applied to the Terrapin attack, the chosen state is the SSH sequence number. It uses an abstract stateful encryption interface, for which we provide pseudocode descriptions for the eight most prominent AEAD modes used in SSH. By varying the SND oracle, we can model ciphertext-only (CO; the Terrapin attack), known-plaintext (KPA), and chosen-plaintext (CPA) attacks. This allows us to establish concrete bounds on the security of the AEAD modes. We find that all three Encrypt-then-MAC (EtM) modes and ChaCha20-Poly1305 in SSH are insecure in the CO model. AES-GCM is the only cipher secure in all three model variants. Going beyond Terrapin, we show that Encrypt-and-MAC (EaM) with a CBC cipher is secure, even in the KPA model. In particular, we describe a novel BEAST-like chosen-plaintext attack on the channel integrity of EaM-CBC, which separates the KPA and CPA models for this scheme.

Concept drift, driven by the rapid evolution of Android malware, severely degrades the performance of machine learning detectors. Current adaptation strategies are often reactive, responding only after performance has dropped and imposing a significant manual annotation burden, or they are proactive but rely on unstable adversarial training and incomplete, single-level graph representations. To overcome these limitations, we propose HYDRA (Hybrid Drift Adaptation), a proactive adaptation framework that learns drift-invariant representations from hierarchically structured data. HYDRA first models applications using a hybrid graph structure, combining fine-grained Control Flow Graphs (CFGs) and coarse-grained Function Call Graphs (FCGs) to capture comprehensive behavioral patterns. It then introduces a novel cross-domain contrastive learning objective that aligns historical (source) and new (target) data distributions. By generating pseudo-labels for unlabeled target samples, our method pulls representations of semantically similar applications together, regardless of their domain, within a single, stable optimization process. This approach unifies feature learning and domain alignment, eliminating the need for complex adversarial objectives. Extensive experiments on large-scale, time-ordered malware datasets demonstrate that HYDRA achieves substantially lower False Negative and False Positive Rates than state-of-the-art baselines while requiring up to 87.5% fewer labeled samples. Our work thus offers a robust and efficient solution to combat concept drift in security applications.

This paper presents a controlled post-alert incident orchestration and response subsystem for educational information systems. The architecture separates deterministic classification, contextual analysis, human approval, and technical execution. A Rule Engine determines severity and selects the playbook, while Static RAG and a local large language model provide advisory content under Validator, Guardrail, Output Sanitizer, and Safe Fallback controls. Experiments begin after simulated alerts are stored in Elasticsearch. The Rule Engine matched the predefined routing matrix in all 30 boundary cases. The Durable Queue completed 100 events without duplicate tasks, new failed tasks, or unintended firewall rules. An eight-alert contention experiment preserved the configured limit of one active model request, and 30 sequential measurements showed an overall mean post-alert processing time of approximately 33 seconds. The results demonstrate functional correctness, traceability, controlled recovery, and bounded model integration within the evaluated laboratory scope.

Stateless UTXO-style execution validates transactions using local and referenced data, enabling parallel validation and predictable serialized-size/weight accounting. Multi-step workflows thread state across outputs, and a prepared next-step transaction may become stale if another valid spend confirms first. Explicit state threading therefore shifts consistency maintenance, off-chain tracking, and transaction rebuilding to the protocol boundary, increasing coordination cost and latency. Recursive invariants (RIs), our proposed transaction-level logic and toolchain, address this gap by expressing workflow rules as transaction-level predicates over a transaction's inputs and indexed successor positions referenced by the RI. An accepted transaction realizing such a successor position re-checks the predecessor's RI one step later, carrying the workflow rule forward without shared mutable application state or executable output logic. Thus, multi-step protocol rules preserve validation-time locality and admit explicit cost accounting, while cross-transaction guarantees arise from repeated one-step checking. Not all successor clauses are checkable at validation time, so our small statically typed domain-specific language (DSL) uses three-valued semantics (true, false, unknown) to defer future-dependent obligations until checkable. Co-designed with this DSL, our framework formalizes UTXO validation and ledger extension, identifies the validation-time-evaluable one-step fragment, and proves the deduction system sound w.r.t. the three-valued semantics. We give validation and ledger-extension algorithms for this model. We implement a prototype RI interpreter and benchmarking toolchain for six workloads. Six practice-motivated case studies exhibit roughly linear cumulative validation-cost proxy growth and illustrate staged workflow constraints without preconstructing each successor.

The mobility sector is undergoing a paradigm shift driven by advances in Generative Artificial Intelligence. With a global market valued at approximately 2.9 trillion dollars annually, considering only cars, the integration of these technologies has the potential to impact more than 1.5 billion vehicles worldwide. As Large Language Models (LLMs) are increasingly adopted in mobility, concerns about cybersecurity, privacy, and reliability emerge. Accordingly, this paper surveys current applications and assesses these challenges. Since the European AI Act classifies transportation AI as high risk, we derive nine technical classes from its requirements to assess current research and future deployments. Our findings show that research mainly studies GPT and Llama models (over 50\% of reviewed works) and traffic applications while largely neglecting security, privacy, and reliability. This gap extends to AI Act compliance: among 35 reviewed works, only one includes a partial vulnerability assessment and one a partial risk management system. We identify a clear gap between strong optimization performance and regulatory adherence, suggesting compliance is limited less by technology than by a focus on static performance over lifecycle safety, and underscoring an urgent need for security-by-design in safety-critical intelligent transportation systems.

Cybersecurity literature has extensively documented the operational benefits of artificial intelligence (AI) for threat detection, incident response, and prevention, while raising qualitative concerns about over-automation, algorithmic bias, and analyst-skill erosion. What remains largely absent is a formal, falsifiable model connecting three constructs that recur across this literature: Defense-in-Depth Theory, the Artificial Intelligence Theory of Pattern Recognition, and human-AI collaboration in security operations. This paper develops such a model. We formalize layered defense as a Bernoulli detection cascade in which AI augmentation enters multiplicatively across layers; we formalize each layer's pattern-recognition behavior as a Neyman-Pearson/Bayesian detector with a derived closed-form optimal threshold; and we formalize human-AI triage as a capacity-constrained cascade with an explicit, quantifiable trade-off between detection probability and false-alarm ("alert fatigue") rate. A Monte Carlo/analytical simulation evaluated at illustrative but realistic operating points shows that (i) AI augmentation compounds across defense layers, delivering its largest marginal gains exactly where traditional layering saturates, and (ii) full human review of AI-flagged alerts is not optimal: increasing analyst capacity toward 100% coverage cuts false alarms by roughly 20-fold but simultaneously lowers system-level detection probability, because imperfect analyst accuracy is then applied to every alert rather than a filtered subset. These results give the widely repeated qualitative recommendation of "balanced human-AI collaboration" a precise, testable form and suggest an interior-optimum capacity ratio as a concrete design target for security operations centers (SOCs), including those securing IT/OT-converged critical infrastructure.

DAG-based consensus protocols can achieve great throughput and the optimal three-message-delay limit for n = 3f+1 consensus. While two-delay protocols exist, they pay with reduced resilience (requiring 5f+1-style committees) or rely on fallbacks that sacrifice the DAG's high throughput. This paper introduces Hydrozoan, the first DAG protocol with a dual commit path under a hybrid fault model of f Byzantine and c crashed validators, on n = 3f+c+2p+1 validators. Leaders commit in two message delays whenever at most p validators are faulty, and in three otherwise, with no extra messages, no view changes, and multiple leaders per round. Both paths are evaluated on the same DAG, using a novel graded indirect rule to reconcile them so that every honest validator reaches the same decision. We show that under geo-distributed conditions, which path is faster is a property of geography rather than the protocol, as rounds reaching a remote region cost far more than those that do not. The (f, c, p) knobs place the fast quorum where the deployment requires it, allowing a commit in two message delays. If misconfigured, Hydrozoan can still commit in three message delays: Hydrozoan commits on whichever path fires first. We also present Optimal-Hydrozoan, a variant that tolerates one more fault on the fast path, the first construction to match the known lower bound. The safety and liveness of both protocols are machine-checked in Lean 4. Our geo-distributed evaluation shows that Hydrozoan matches Mysticeti's throughput, commits ~25% faster when the fast quorum fits fast regions, and falls back to three message delays when it does not or past p faults, where existing two-delay protocols stall.

Many Security Operations Centers rely on signature-based Network Intrusion Detection Systems like Suricata, yet detection rule engineering remains understudied. We investigate this process by introducing SuriCap, a platform for rule engineering exercises, and hosting CTF-style workshops where 60 participants, trained MSc students, and experienced SOC professionals, created rules for four scenarios. Participants produced 3146 valid rules, enabling analysis of their methods, performance, and iteration patterns. Surprisingly, prior experience had limited impact on rule quality, suggesting that less experienced engineers can produce rules comparable to experts. We also observed challenges in generalizing rules beyond available tests, underscoring the need for sufficient labeled data. From our study, we identify three phases and a common pattern in rule engineering, offering SOC managers insights to improve their processes and expectations of engineer expertise.

Users commonly combine multiple Low-Rank Adaptation (LoRA) adapters to personalize images with different subjects, styles, and visual attributes. Yet inspecting adapters individually does not establish the safety of their composition. We identify and characterize a pair-conditioned attack in text-to-image diffusion: individually useful and benign-appearing adapters redirect image generation when co-loaded with a specifically matched partner, whose identity serves as the trigger. We introduce LoRango to realize this attack through complementary Signature and Payload adapters. The Signature writes a pair-specific code into intermediate carrier representations, while the Payload uses code-selective responses and opposing signal/reference branches. These branches approximately cancel for standalone adapters and mismatched pairs; matched code-reader alignment breaks cancellation within native GEGLU blocks and releases the programmed action. Both adapters are exported as ordinary static LoRA files compatible with standard loaders, requiring no prompt trigger or base-pipeline modification. LoRango achieves matched-pair attack success rates of 97.9\% on SD v1.5 and 98.7\% on SDXL, compared with 2.8--4.6\% when implanted adapters are loaded individually. Further experiments evaluate pair selectivity, standalone fidelity, robustness to deployment variations, and applicability across denoiser architectures. These findings show that individual-adapter inspection is insufficient to assess the security of multi-LoRA personalization and motivate auditing adapter compositions.

The use of malicious domains is central to cyberattacks such as phishing, malware distribution, impersonation, and fraudulent transactions. Because domains are inexpensive to register and easy to deploy at scale, they remain one of the most common and damaging tools used in cybercrime across industries. Proactive detection is essential to reducing this window of vulnerability and preventing harm to users. In this work, we propose COBRA: a content-agnostic, registration-time detection framework for identifying and analyzing suspicious domains from day zero. Our approach does not rely on any content-based features, allowing us to classify a domain even before it is populated with content. We analyze the names of newly registered domains and employ a clustering technique to group them based on lexical and structural similarity. We evaluate our methodology using real-world data consisting of 1.5M newly created domains, demonstrating that COBRA detects suspicious domains with a precision of 98.5%, identifying more than 47K distinct newly registered suspicious domains. Furthermore, our results show that domain-name clustering enables accurate early detection, allowing us to identify 80% of suspicious or malicious domains earlier than one of the most widely used threat-intelligence services, which in some cases may require up to 7 days.

Trapdoor claw-free functions (TCFs) underpin much of classical-quantum cryptographic interaction, yet every TCF-based protocol states its guarantees relative to an honestly generated key. We give a family-agnostic abstraction of key certification for (noisy) TCF constructions, built on two notions: a certifiable key relation, an NP relation capturing a family's honest keys with witnesses recoverable from the trapdoor; and certified key generation, which emits with each key a certificate of membership satisfying completeness, certificate soundness with extractability, and key privacy. We instantiate certifiable key relations for different constructions, each met generically by a zero-knowledge argument of knowledge for the relation. As our main application, a generic compiler turns any TCF-based proof of quantumness into a zero-knowledge one, with each security property following from its counterpart in the certification scheme. Finally, we delimit the primitive's reach: for protocols resting on injective invariance, an accepting certificate is itself a family distinguisher, leaking exactly the bit such protocols must hide.