Papers

System-generated logs underpin security monitoring, yet their rigid template-based format hinders both automated analysis and human comprehension. We present NLLog (Natural-Language Log), a lightweight pipeline that deterministically rewrites parsed templates into WHO-WHAT-SEVERITY sentences, pools them with term-frequency-inverse-document-frequency weighting, classifies sessions with tree ensembles, and back-projects evidence with TreeSHAP for analyst review. On Hadoop Distributed File System (HDFS) and Blue Gene/L (BGL) corpora, NLLog exceeds two reproduced matched-protocol baselines; across HDFS, BGL, and the AIT Alert Data Set, it sustains low false-positive rates with commodity-hardware latency suitable for security operations center triage. Coverage, sparse-versus-dense, faithfulness, and adversarial ablations show that fallback sufficiency is corpus-dependent, that an enrollment-time coverage check can surface refinement requirements before deployment, and that an auditable deterministic rewrite combined with lightweight dense encoding provides a measurable representation layer for log-anomaly detection and triage.

LLM post-training proceeds through multiple stages, e.g., supervised fine-tuning (SFT) followed by reinforcement learning from human feedback (RLHF) or direct preference optimization (DPO), where each stage draws data from different, potentially untrusted sources. Existing literature assumes data poisoning attacks may occur at each training stage, but neglects the possibility of multiple attackers. To study the trustworthiness of the entire post-training pipeline, we propose the threat model of sequential data poisoning, where multiple adversaries separately poison the SFT and preference datasets. Under this threat model, we identify the single-attacker illusion: each adversary, evaluated in isolation, appears to pose a negligible threat. Yet when adversaries collaborate across stages, the true vulnerability is revealed. In the SFT $\to$ DPO pipeline, their contributions are additive: splitting a fixed poison budget across stages outperforms concentrating it in either stage alone. In the SFT $\to$ PPO pipeline, their contributions are complementary: neither SFT nor reward model poisoning succeeds individually, yet their combination does. These findings show that security analyses of individual post-training stages systematically underestimate compound vulnerabilities that emerge only from their interaction. Code is available at https://github.com/jcksanderson/sequential-poisoning.

Trusted Execution Environments (TEEs) have emerged as a critical technology for safeguarding sensitive data and ensuring code integrity in modern computing systems. However, relying on a single TEE implementation makes systems vulnerable to a central point of attack. Building distributed-trust systems leveraging heterogeneous TEEs helps disperse trust but still faces threats from centralized management and adaptive mobile adversaries. To address these challenges, this paper introduces TeeDAO, a novel three-layer framework that automatically organizes multiple heterogeneous TEE instances and provides unified interfaces to support diverse applications, while ensuring long-term guarantees of availability, integrity, and confidentiality. TeeDAO couples BFT-ordered governance with heterogeneity-aware Distributed Proactive Secret Sharing (DPSS) and Secure Multi-Party Computation (MPC) so that attestation-driven committee changes are consistently reflected in secret recovery, resharing, and computation across a dynamic committee of heterogeneous TEEs. We implement a prototype of TeeDAO, integrating COBRA's DPSS scheme with the HotStuff BFT consensus protocol, and adapt it for Intel SGX, TDX, and Hygon CSV. Evaluations demonstrate that TeeDAO achieves up to 1.8x higher key-value store throughput in a large cluster with 61 nodes compared to state-of-the-art systems, efficient autonomous management, and minimal computation overhead (<18%) for multi-party computation tasks.

Low-Earth Orbit (LEO) mega-constellations such as Starlink by SpaceX and Kuiper by Amazon rely on optical Inter-Satellite Links (ISLs) for autonomous mesh routing to provide low-latency telecommunication, Internet of Things (IoT), and security services globally. As commercial operators and governments deploy increasingly dense constellations and form multi-operator peering coalitions, ISL integrity becomes critical to both commercial availability and national security. However, there is a lack of real-world data for LEO constellations and existing real-time security approaches focus strictly on physical layer security, leaving blind spots in the coverage of network-layer and composite attacks. In this paper, we present a cross-layer, lightweight behavioral fingerprinting framework that fuses onboard physical-layer measurements with network-layer data to detect anomalies at low computational overhead. We construct an orbital simulation covering the first shells of Starlink (1,584 satellites), Kuiper (1,156 satellites), and a joint multi-operator peering scenario (2,740 satellites), injecting ten attack types that span spoofing, traffic manipulation, and routing subversion at varying severity. We evaluate three unsupervised, per-satellite detectors among which our Mahalanobis-distance-based detector achieves 99.5% recall on Starlink, 99.4% on Kuiper, and 94.8% on the multi-operator constellation, while maintaining False Positive Rates (FPR) below 0.7%. Our results demonstrate that cross-layer feature fusion is not only necessary for comprehensive security of LEO constellations but highly cost-effective for large-scale networks while fitting into the strict onboard energy budgets of resource-constrained satellites.

Trusted Execution Environments (TEEs)-aided federated learning protocols emerge as promising solutions to counter server-side adversaries and ensure the trustworthiness of the server. In this paper, we dissect existing protocols and demonstrate that server-side adversaries can still manipulate client selection and replay aggregation to compromise system robustness and privacy, by exploiting TEE limitations, i.e., state rollback and I/O manipulation. To this end, we present DIST-FL, a distributed system of servers guarded by multiple TEEs forming an append-only ledger for privacy-preserved, robust FL aggregation. Specifically, DIST-FL ensures operation linearizability to thwart state rollback attacks and incorporates inputs from reliable servers to mitigate I/O manipulation threats. We implement DIST-FL and conduct evaluations in WAN settings. Experimental results demonstrate that DIST-FL can effectively counter the proposed attacks and match the single-TEE's performance while offering a 6x throughput boost over its counterparts, leveraging TEE's computational advantages.

Confidential blockchains leveraging Trusted Execution Environments (TEEs) have garnered extensive attention for transaction confidentiality. In this paper, we first taxonomize two classes of attacks against confidential blockchains, i.e., execution-inference and execution-replay attacks, which exploit TEEs' long-lasting side-channel and state-continuity issues to compromise the confidentiality of existing consortium blockchains. Then, we present ODYSSEY, a confidential blockchain that efficiently mitigates these attacks. The core innovations of ODYSSEY are the following: (1) Its delegation model: clients delegate transaction execution to their designated trustees, while other participants synchronize only the execution results, which significantly reduces the attack surface while preserving confidentiality and system performance. (2) Two novel techniques to improve ODYSSEY's efficiency and security: location-aware concurrent execution and delegation failure handler. Finally, we develop a prototype of ODYSSEY on FISCO BCOS, an enterprise-grade consortium blockchain platform. We have conducted various experiments, and our evaluation results show that in a WAN environment with 3 nodes, ODYSSEY can achieve about 4k throughput while keeping latency as low as 0.4-0.5s.

Pearl, a Layer-1 blockchain with high-profile AI industry endorsements, markets its Proof-of-Useful-Work (PoUW) protocol as simultaneously securing the network and performing AI inference. We present the first systematic empirical measurement of a deployed PoUW system, finding that Pearl's 24 EH/s network – representing approximately 320,000 GPU-equivalents consuming an estimated 112 MW – produces zero useful AI computation. Budget GPU rental prices rose 38% and utilization surged from 57% to 94% following the mining software's public release, displacing legitimate research workloads. Our measurements span five dimensions: (1) network composition analysis of 8,012 workers shows all have inference-capable hardware, yet the dominant mining software contains no inference code; (2) the verification protocol accepts random matrices by design, confirmed by 44 pool-accepted shares from our open-source miner across NVIDIA, AMD, CPU, and Apple Silicon hardware; (3) statistical distribution checks are trivially defeated by adversarial Gaussian sampling; (4) mining economics are marginal at current PRL prices ($0.76), with ROI ranging from -1% to +67% depending on GPU tier – near breakeven for most hardware; and (5) the mining computation is commodity integer arithmetic portable to any hardware platform, offering no vendor lock-in. These findings quantify the verifiability-usefulness tension identified theoretically by Leinweber et al., providing concrete measurements of its magnitude and economic consequences in a deployed system.

The Model Context Protocol (MCP) has emerged as a critical standard empowering Large Language Models (LLMs) to utilize external tools. In this ecosystem, LLMs rely on natural language descriptions provided by MCP servers to select and execute functions. This interaction implicitly assumes that tool descriptions faithfully reflect their underlying implementations, while this assumption is not mandatorily verified in practice. As a result, MCP deployments may suffer from a problem named Description-Code Inconsistency (DCI), where a tool's description of its capabilities and security boundaries is not consistent with what the code actually does. In this paper, we present a comprehensive study of DCI in real-world MCP servers. We formally define the problem and propose a comprehensive taxonomy spanning functionality inconsistencies and undeclared side effects. Guided by this taxonomy, we develop DCIChecker, an automated framework that combines structure-aware static analysis with the Direct-Reverse-Arbitration prompting method to cross-validate tool descriptions against actual code implementations. We apply this framework to a large-scale dataset comprising 19,200 description-code pairs extracted from 2,214 real-world MCP servers. Our measurement reveals that DCI is widespread, with 9.93% of these pairs exhibiting inconsistencies. We further demonstrate that DCI creates a critical defense blind spot, facilitating varied risks from operational failures to stealthy malicious behaviors. Finally, we propose mitigation strategies to enforce semantic consistency and enhance the reliability of the emerging agentic ecosystem.

Chain-of-thought (CoT) answer-hijack templates can flip the final numeric answer of a 7B-8B language model on GSM8K or MATH-500 even when the visible reasoning trace looks fluent. Activation patching is the standard probe for locating where this hijack can be undone, and a successful clean-source patch is often read as evidence that the patched activation carries the recovered content. We show that this reading is unsound: clean-only localization profiles (peak, spread, thresholded band) underidentify the frozen-hook source contrast, and the clean-only profile is an intervention map, not a mediation certificate. We then construct an audit that turns each candidate patch into a source-control certificate with a pre-registered Type-I guarantee. The certificate runs in three stages: SELECT (clean-source band sweep with permutation calibration and held-out validation), FREEZE (lock the hook), and AUDIT (paired-bootstrap source contrasts at the frozen hook). It emits an incorrect mechanism label with probability at most alpha = alpha_sel + alpha_audit under sample-split disjointness. A matching-rate sample-complexity theorem (n_star = Theta(Delta^{-2} log(1/alpha))) bounds the audit cost. On Qwen2.5-7B and Llama3-8B, three few-shot/puzzle cells pass confirmatory K=1 localization with held-out gaps +32.6, +45.1, +17.7; fixed-hook reruns recover 47.0% (Qwen-puzzle) and 39.0% (Llama3-puzzle) at n=100; frozen MATH-500 transfer recovers 26.0%. After audit, Llama3-PZ and Qwen-PZ are identity-light with moderate magnitude (Qwen-PZ also layer-sensitive); Llama3-FS is a single-seed moderate-positive candidate (multi-seed replication queued); Qwen-FS is exploratory non-separation with a layer-sensitive flag. The method is a diagnostic auditing protocol, not an adaptive safety defense.

Secure implementation of post-quantum cryptography (PQC) requires attention to cryptographic mechanisms, software integration, developer capability, and organisational support. Understanding how available approaches address these requirements is important for preparing software systems for quantum threats. This Systematisation of Knowledge (SoK) synthesises 30 publications and analyses PQC implementation approaches and challenges using a Human, Organisational, and Technological (HOT) perspective. We identify four approach categories: guidelines, frameworks, tools and libraries, and educational interventions. Technological support receives greater representation in the extracted mapping, while no approach is classified primarily as organisational, despite secondary organisational contributions in some approaches. The challenge synthesis identifies five layers covering implementation security, system integration and lifecycle, tooling, organisational governance, and human factors. Together, these findings highlight a difference between the primary support functions of the mapped approaches and the breadth of the reported implementation challenges. We propose PQC-HOT, an evidence-informed analytical framework connecting implementation tasks with technical resources, practitioner capabilities, and organisational arrangements. We derive research priorities and engineering implications to guide framework evaluation and support secure, maintainable PQC-enabled software.

This paper presents TeleHunt, a framework and tool for evaluating the effectiveness of different strategies to discover cybercriminal communities on Telegram. TeleHunt employs a set of reference-driven snowballing strategies, integrating message-level classification, contextual filtering, and market-segment labeling. Using open- and dark-web seeds, we systematically evaluate how seed source, pointer type, and exploration strategy influence discovery outcomes in three dimensions: efficiency, accessibility, and rediscovery. Our work provides (i) a modular cybercrime content discovery pipeline, (ii) the first systematic comparison of Telegram discovery strategies with an empirical characterization of market-segment accessibility, and (iii) a labeled dataset of over 172 million messages from 6,022 Telegram communities.

Cyber threat signals are fragmented across multiple social media platforms, yet no existing approach has fully automated their integration into actionable threat intelligence (TI) reports. We present TIBlender, a multi-agent system that monitors four platforms (X, Reddit, Telegram, and Discord) and produces structured TI reports via role-specialized LLM agents. These agents conduct multi-perspective investigations, tracing chains of evidence to uncover related Indicators of Compromise (IoCs) via collaborative, evidence-backed analysis. In a real-world deployment, TIBlender detected emerging threats across all four threat categories ahead of public feeds, including in-the-wild exploitation ahead of public vulnerability registries; the majority of its IoCs were absent from each evaluated feed. Quantitative evaluation confirms that each platform contributes unique threat information unavailable from the others, and that excluding any single platform results in substantial loss of reports in specific threat categories. Under identical single-platform input conditions, TIBlender's direct extraction produced substantially different IoC sets from the corresponding platform-specific baselines, while multi-perspective investigation further enriched the resulting intelligence. These results establish cross-platform social media monitoring as an effective and scalable early-warning layer for operational TI pipelines.

Confidential Virtual Machines (CVMs), such as AMD SEV-SNP, enable cloud tenants to run security-sensitive workloads, but tenants can rely on the execution of these workloads only when they can trust the CVM. This trust requires continuous integrity assurance from CVM launch to the current runtime state, including initial trust establishment at launch and subsequent runtime integrity assurance. Existing works help establish launch-time trust and protect parts of runtime integrity, but they do not fully address the integrity of file-backed user-space executable objects, such as main executables, program interpreters, and dynamically loaded shared objects, that may be loaded or mapped dynamically during execution inside CVMs. In this paper, we propose Privilege-Separated User-space Integrity Enforcement (PS-UIE), an approach for enforcing the integrity of user-space executable objects inside AMD SEV-SNP-based CVMs. PS-UIE consists of a privilege-separated architecture and three mechanisms. The architecture separates the authority for integrity measurement and enforcement from the measured targets by placing it in a higher-privileged protected domain. Built on this architecture, PS-UIE provides policy lifecycle management, execution-time integrity enforcement, and evidence export and verification mechanisms. It enables policy-controlled integrity measurement and enforcement for user-space executable objects and generates verifiable runtime evidence. We implement PS-UIE on an AMD SEV-SNP platform. The security analysis and performance evaluation show that PS-UIE enforces the integrity of user-space executable objects on the covered execute-permission grant paths and provides verifiable runtime evidence while incurring acceptable overhead.

Watermarking methods for language models have been studied extensively in the autoregressive setting, where tokens are generated sequentially. These works largely focus on local-context schemes that perturb the next token's distribution as a function of its preceding tokens. In diffusion language models, distributions over many unresolved positions are jointly sampled, allowing additive statistics of the entire sequence to be tractable during generation. We propose a watermark for masked diffusion language models that controls a global, vector-valued sketch representation of the text. Compared to context-dependent watermarking, the sketch formulation decouples detection from the local contexts seen during generation, resulting in an order-agnostic statistic and a watermarking rule which does not manifest as a simple token bias. We analyze the distortion, soundness, and robustness properties of the method.

AI has the potential to transform cybersecurity by enabling systems that can autonomously detect, analyze, and remediate software vulnerabilities. However, existing cybersecurity evaluations of AI systems are limited in scale or scope, and fail to capture the end-to-end lifecycle of real-world software vulnerability discovery and remediation. To address this gap, we propose CyberGym-E2E, a large-scale and realistic end-to-end cybersecurity benchmark that comprehensively evaluates AI agents' abilities across the full lifecycle of vulnerability discovery, PoC generation, and patch generation. CyberGym-E2E is comprehensive and scalable, as we build an automated, agent-enhanced pipeline for transforming open-source vulnerability data into realistic evaluation environments. Currently, the benchmark consists of 920 real-world vulnerabilities across 139 different open-source projects.

Language model parameters are known to impose unique (to each model) geometric constraints on their logit outputs, which serves as a signature that identifies the model, but also leaks the model's final layer parameters when an API distributes logits. We investigate more restrictive APIs that expose token rankings (i.e., their ordering by probability, but not the probability values) and find that rankings also constitute a signature: every model has a unique set of feasible top-$k$ rankings for sufficiently large $k$. Furthermore, the ranking signature is the first known (polynomially) unforgeable signature, since finding a model with the same set of feasible rankings is NP-hard. On the security front, we find that token rankings are already sufficient to approximately steal the final layer of the model, similar to logits, though the approximation is too coarse to forge the signature, and can be effectively countered by restricting the API to top-$k$ tokens with sufficiently small $k$. Since the top-$k$ required to present the model signature is generally smaller than the $k$ required to prevent stealing, it is possible for an API to present an unforgeable signature without leaking model parameters.

Black-box tests for Fujisaki-Okamoto decapsulation observe the sampled execution seen by the harness, whereas the reencryption computation itself is visible only through the values that reach final key derivation. We study confirmation-code-augmented KEM variants under an honest-reference harness in which the reference encapsulation fixes a hidden final-key point $\langle good,B,W\rangle$, with $W$ the confirmation witness. For a $q$-localized system under test, acceptance is bounded by honest correctness error, adversarial aliasing, final-key freshness defects, a hit on the localized suffix list $Q_G(B)$, and $2^{-\kappa}$. A one-query construction from any predictor of $W$ matches this bound up to the fresh-key coincidence term, so the list-hit event is the black-box obstruction measured by the harness. The list-hit term is bounded either by a cUP-faithful harness certificate, which transfers source confirmation-code unpredictability with a $q$-loss, or by an average conditional min-entropy bound, with separate RawEnt and TailEnt hypotheses for short diagnostic and truncation-tail codes. The same model proves a dependency-cone lower bound for non-certification claims. When the black-box observation of an honest-support harness factors through the confirmation-observable final-key target, every operation outside the support-active cone has a coupled erasure implementation with the same transcript distribution; over any implementation class containing that erasure, soundness and completeness errors of an execution certifier satisfy $\alpha+\beta\ge 1$. The ML-KEM and HQC case studies distinguish theorem-covered positive rows, finite-catalog artifact rows, and non-certification rows that carry a cone-inactivity certificate. The security of the standard KEM lines is the construction-level security supplied by the cited source analyses.

Multi-step agentic retrieval-augmented generation (RAG) pipelines have demonstrated significant capability for complex reasoning tasks, yet remain vulnerable to a class of failure that existing hallucination detection mechanisms systematically miss: cascading hallucination, where errors introduced at early pipeline stages propagate and amplify across successive reasoning steps, producing confident but factually incorrect final outputs. To address this vulnerability, we formalize cascading hallucination as a distinct failure mode in agentic RAG systems, present a four-type taxonomy of cascade patterns, and introduce CHARM (Cascading Hallucination Aware Resolution and Mitigation), an architectural framework for detecting and interrupting error propagation in multi-step reasoning pipelines. CHARM comprises four components - stage-level fact verification, cross-stage consistency tracking, confidence propagation monitoring, and cascade resolution triggering - that operate alongside standard agentic RAG pipelines without requiring architectural replacement. We evaluate CHARM on HotpotQA, MuSiQue, 2WikiMultiHopQA, and a custom adversarial dataset across LangChain agentic pipeline configurations, achieving an 89.4% cascade detection rate with a 5.3% false positive rate and 215 ms +/- 18 ms average latency overhead per stage, achieving an error propagation reduction of 82.1%, compared to 18.5% for output-level detectors. Component ablations confirm that each detection module contributes meaningfully to overall cascade coverage. CHARM integrates with human-in-the-loop oversight frameworks to provide a complete reliability and governance stack for production agentic AI deployment.

Modern agentic systems fundamentally reshape the security boundary of LLMs by introducing persistent system state including memories, filesystems, tools, and other long-lived contextual artifacts that survives across sessions. As external information crosses this boundary and becomes part of persistent agent state, malicious instructions are no longer confined to a single interaction, but can silently persist and influence future executions long after the original attacker interaction has ended. We introduce Cross-Session Stored Prompt Injection, a new threat vector inspired by stored cross-site scripting that redefines prompt injection for agentic systems by extending its threat model across both time, where attacks persist and activate across sessions, and space, where adversarial instructions propagate beyond the immediate prompt into persistent system state. To systematically characterize this emerging threat, we formalize the lifecycle of cross-session stored prompt injection, develop a taxonomy of persistence channels and incorporation mechanisms, and build a sandbox toolkit for evaluation. Our findings suggest that the fundamental challenge of agent security is not merely filtering untrusted inputs, but governing how external information acquires authority as it crosses persistent system boundaries. We hope this work motivates a broader shift from interaction-centric security toward state-centric security, making the secure management of persistent agent state a first-class security principle for the agentic era.

We present Pepper, a high-bandwidth anonymous broadcast protocol that provides cryptographic sender anonymity against global adversaries. Pepper builds on a two-server DC-net architecture but introduces three key innovations: a self-contained anonymous registration subprotocol using verifiable distributed point functions, support for batch messaging via distributed multi-point functions, and a lightweight access control mechanism based on secret-shared proofs. Unlike prior systems, Pepper eliminates the need for external dialing services and allows each broadcaster to send multiple messages per epoch with a single audit, significantly improving throughput for large data transfers. Our implementation demonstrates that Pepper achieves millisecond-level registration audits, scales efficiently to thousands of channels, and delivers 1.2–20$\times$ higher effective messaging rates than state-of-the-art alternatives. Furthermore, Pepper is designed for practical deployment, with natural compatibility for co-deployment alongside Tor and federated social networks.

In the paradigm of decentralized learning, a group of agents collaborate to train a global model using distributed datasets without a central server. Although the power of collaboration has been verified by many state-of-the-art studies, it entails extensive gradient information exchanging among the agents and thus induces high risk of privacy leakage for the individual agents. Moreover, in real-world applications, the training data are usually non-identically and independently distributed across the agents, inducing more challenges to enable privacy-preserved decentralized learning. To address these issues, we propose a privacy-preserved decentralized learning algorithm with non-IID data, DPDL, which leverages the notion of Differential Privacy (DP) in cross-gradient aggregation through a similarity-based calibration technique. Specifically, in each round, each agent perturbs the cross-gradients (i.e., the derivatives of its neighbors' local model in its private local data) by Gaussian noise mechanism before sharing them with its neighbors; it then adopt cosine similarity to calibrate the received perturbed cross-gradients such that the aggregation of the calibrated cross-gradients can be utilized to effectively update local model in a momentum-like manner. Our rigorous theoretical analysis not only reveals the minimum noise level required to achieve a specific level of privacy preservation, but also illustrates that our algorithm still achieves a linear speedup in training with non-IID data. We finally conduct extensive experiments on real-world dataset to validate the effectiveness of our algorithm in defending privacy attacks and in training accurate models.

Federated Learning (FL) has emerged as an effective paradigm for collaborative intelligence while preserving data privacy. However, data heterogeneity arising from non-IID distributions and decentralized security threats remain significant challenges, particularly in resource-constrained enterprise environments. This paper presents TITAN-FedAnil+, a Trust-Based Adaptive Network for blockchain-enabled federated learning in intelligent enterprises. The proposed framework introduces affinity propagation-based adaptive clustered aggregation to identify and filter malicious updates without requiring prior knowledge of the number of attackers. In addition, GPU-accelerated vectorization is employed to improve computational efficiency, while a signed state jump mechanism enables lightweight blockchain resynchronization. Experimental results demonstrate substantial reductions in memory overhead, achieving up to 81% savings across 50 communication rounds on constrained 8 GB edge devices compared with the baseline framework. The results indicate that TITAN-FedAnil+ effectively improves robustness, scalability, and resource efficiency for secure federated learning deployments in intelligent enterprise environments.

Machine learning's reliance on sensitive data necessitates privacy-preserving techniques like Differentially Private Stochastic Gradient Descent (DPSGD). However, DPSGD suffers from substantial utility degradation and slow convergence due to gradient clipping and noise injection. Prior works have attempted to improve DPSGD from various perspectives; notably, the Differentially Private Selective Update and Release (DPSUR) algorithm has achieved remarkable model utility. However, the privacy accounting in DPSUR overlooks the variation in sampling probability introduced by the selective release mechanism, which compromises the rigor of its privacy guarantees. To address these limitations, we re-evaluate the privacy analysis of the selective release mechanism and propose a novel algorithm: Differentially Private Selective Release based on Clipped Gradients (DPSR-CG). Through a rigorous, newly derived privacy analysis and extensive experiments on multiple datasets (MNIST, CIFAR-10, IMDB, and FMNIST), we demonstrate that our DPSR-CG mechanism maintains strict privacy guarantees while achieving exceptional model performance.

Privacy-sensitive and distributed characteristics of multi-center medical data bring severe obstacles to centralized modeling for accurate early prediction of sepsis. Federated learning (FL) has attracted growing attention as a promising framework for collaborative model development, as it allows multiple institutions to jointly train predictive models without directly sharing or centralizing raw data. Nevertheless, its practical performance, robustness, and privacy-preserving benefits remain insufficiently evaluated using real-world clinical datasets. To bridge this gap, this study systematically examines the application of federated learning to multi-center sepsis prediction. The experimental dataset consists of 648 clinically screened samples collected from three tertiary hospitals in China, with rigorous inclusion and exclusion criteria. We establish a centralized training paradigm as the performance baseline, and then implement a horizontal federated learning framework for distributed collaborative modeling. Extensive experimental results demonstrate that the federated learning-based model achieves highly comparable prediction accuracy to the centralized counterpart, while fundamentally avoiding privacy leakage. Further privacy security analysis verifies that malicious attackers cannot reconstruct the original patient data from the transmitted model parameters, indicating strong resistance against data reconstruction attacks. This work not only validates the practicality and security of federated learning in clinical sepsis prediction, but also provides a reliable and feasible solution for privacy-preserving multi-center medical collaboration.

Memory is a core component of AI agents, enabling them to accumulate knowledge across interactions and improve performance. However, persistent memory introduces the risk of memory poisoning, where a single adversarial memory write can exert long-term influence over agent behavior. We present a systematic study of memory poisoning in LLM-based agents. We identify four memory write channels and nine structural vulnerabilities in model capabilities, system prompt design, and agent system architecture that make these channels exploitable. Based on these vulnerabilities, we develop a taxonomy of six classes of memory poisoning attacks. Furthermore, we design MPBench – a benchmark for evaluating memory poisoning attacks, and show that agents designed to write and retrieve memory more aggressively are more exploitable. We also show that existing prompt injection defenses fail to cover memory poisoning attacks. Our findings provide a foundation for understanding and mitigating memory poisoning attacks against AI agents.

Deep neural networks are increasingly deployed across heterogeneous and partially untrusted environments, where models are distributed through cloud storage, CI/CD pipelines, containerized services, and edge execution platforms. This broad deployment landscape exposes model parameters to various integrity risks. Unlike input-space adversarial attacks, parameter attacks directly tamper with the model's internal parameters and persist across all subsequent inferences. Existing defenses either require retraining, incur significant accuracy degradation, or are limited to specific attack classes. However, in real-world deployment scenarios, the forms of parameter attacks are often unpredictable. To address this challenge, we present ParDef, a generalized defense for deep neural networks against diverse types of parameter attacks. ParDef integrates keyed channel reparameterization, which obscures sensitive parameter directions, QC-LDPC quantization, which embeds redundancy and supports error correction, and adaptive robust inference, which stabilizes predictions under uncertainty. Our evaluation on CIFAR-10, CIFAR-100, and Tiny-ImageNet using ResNet and VGG models demonstrates that ParDef consistently reduces attack success rates across different parameter attacks while maintaining high model performance and incurring only moderate deployment overhead. These results highlight that ParDef is a practical and generalized defense for DNN deployments.

StarkWare's S-two prover provides an efficient means for establishing, on blockchain, that a program written in the Cairo virtual machine language runs to completion. The latter claim is encoded by an algebraic intermediate representation (AIR) that captures the semantics of the Cairo language. The AIR asserts the existence of tables of values from a finite field satisfying certain algebraic constraints. A cryptographic interactive proof system, circle STARK, provides an efficiently-checked certificate that the AIR is satisfied. We describe our verification, using the Lean 4 proof assistant, that the AIR encoding is sound, which is to say, the satisfiability of the AIR implies the computational claim.

Computer-Use Agents (CUAs) are increasingly deployed in dynamic interactive environments, creating a growing need for continual skill learning during interaction. Recent approaches address this challenge by learning reusable skills from successful trajectories. However, these skill learning methods largely assume static and safe environments, overlooking risks from adversarial interactions (e.g., prompt injections) and environmental dynamics (e.g., pop-ups). In dynamic settings, such assumptions can lead to risky skill learning and brittle execution, undermining the reliability of CUAs. This raises the question: how can CUAs learn and use skills safely in dynamic environments? To address this problem, we propose SkillHarness, a framework for safe skill harnessing in dynamic environments. SkillHarness moves beyond static skill abstractions by modeling skill learning and utilization as a safety-constrained interaction process. Specifically, we introduce the skill boundary that leverages multi-source supervision signals to identify safe skills from interaction trajectories, and construct self-improving safety constraints throughout the skill lifecycle. In addition, SkillHarness introduces selective skill reuse, where tasks are guided to decompose according to context and completed through the selective activation of skill subsets. Our experiments demonstrate that SkillHarness significantly reduces the unsafe rate of learned skills by 57.1% and consistently improves execution stability under dynamic environmental changes, outperforming existing baselines.

We present Safecloud, a distributed, encrypted, self-pricing storage and streaming network whose storage and routing nodes never see plaintext and never hold keys. Each file is split into chunks, encrypted on the owner's device, and distributed across Drops (browser tabs storing ciphertext in IndexedDB) and Jets (federated routing servers). Only the owner, or an authorised grantee, can decrypt. We make five contributions: (1) A one-root key hierarchy: every key derives deterministically from a single root via HKDF, and owner and range-scoped grantee derive identical chunk keys (derivation agreement); a subtree key derives its range and nothing else (delegation containment). (2) Convergent content addressing: identical content yields identical ciphertext and identifiers, enabling deduplication without plaintext exposure, with identifiers binding authenticated ciphertext so a keyless Drop verifies integrity (blind verifiability). (3) Three parallel trees over one navigation path (Merkle for integrity, key-derivation for confidentiality, access for authorisation), with sound Merkle-verified retrieval. (4) The key tree doubles as a streaming index: a player derives each segment key in O(1), seeking by derivation, while parallel tracks (video, audio, captions) are independent subtrees unlockable per-track and per-segment, a combination we believe no prior encrypted-storage network offers. (5) Jets and Drops earn Safebux verifiably, kept honest by a one-signature proof-of-storage challenge under chilling-effect Proof-of-Corruption, a zero-sum economy that is significantly cheaper than Filecoin's proof-of-replication sealing (which is slow and provides no confidentiality). We give the architecture, cryptographic construction, a threat model, and an open-source reference implementation, stating precisely what is implemented versus designed.

Multiplayer Online Games have become a multibillion dollar industry in the entertainment sector. However, the presence of cheaters undermines the experience of honest players and devalues the effort of game developers, as it directly affects player retention, competitive integrity, the legitimacy and trustworthiness of a game, and most importantly the overall revenue streams. Among various cheating techniques, visual aimbots represent an emerging threat. They use computer vision models to detect opponents from client screen captures rather than accessing game memory, making them completely undetectable by commercial kernel level anti cheat solutions. In this paper, we introduce PATCH, a novel proactive defense strategy that deploys adversarial patches as in game honeytokens to mitigate the presence of visual aimbot cheaters. Our approach centers on deliberately triggering the cheaters' object detection model, enabling either direct detection, or rendering the game unplayable for the cheater via patch flooding on their viewport. We evaluate our approach on various criteria; analyzing the effectiveness of different patch sizes, scalability of patches to different screen resolutions, efficacy against diverse visual aimbot cheat configurations and also explore various YOLO models to assess patch transferability. Evaluation on a custom Unreal Engine game demonstrates over 90 percent detection rate in white box scenarios for almost all patch sizes, and reaches 60 to 90 percent cross model transferability with larger patches. We further validate our approach on Fortnite, a commercial MOG, demonstrating real world applicability.