Sharif Abuadbba

Shuo Wang, Sharif Abuadbba, Sidharth Agarwal et al.

Existing integrity verification approaches for deep models are designed for private verification (i.e., assuming the service provider is honest, with white-box access to model parameters). However, private verification approaches do not allow model users to verify the model at run-time. Instead, they must trust the service provider, who may tamper with the verification results. In contrast, a public verification approach that considers the possibility of dishonest service providers can benefit a wider range of users. In this paper, we propose PublicCheck, a practical public integrity verification solution for services of run-time deep models. PublicCheck considers dishonest service providers, and overcomes public verification challenges of being lightweight, providing anti-counterfeiting protection, and having fingerprinting samples that appear smooth. To capture and fingerprint the inherent prediction behaviors of a run-time model, PublicCheck generates smoothly transformed and augmented encysted samples that are enclosed around the model's decision boundary while ensuring that the verification queries are indistinguishable from normal queries. PublicCheck is also applicable when knowledge of the target model is limited (e.g., with no knowledge of gradients or model parameters). A thorough evaluation of PublicCheck demonstrates the strong capability for model integrity breach detection (100% detection accuracy with less than 10 black-box API queries) against various model integrity attacks and model compression attacks. PublicCheck also demonstrates the smooth appearance, feasibility, and efficiency of generating a plethora of encysted samples for fingerprinting.

Ee Wei Seah, Yongsen Zheng, Naga Nikshith et al.

The rapid rise of autonomous AI systems and advancements in agent capabilities are introducing new risks due to reduced oversight of real-world interactions. Yet agent testing remains nascent and is still a developing science. As AI agents begin to be deployed globally, it is important that they handle different languages and cultures accurately and securely. To address this, participants from The International Network for Advanced AI Measurement, Evaluation and Science, including representatives from Singapore, Japan, Australia, Canada, the European Commission, France, Kenya, South Korea, and the United Kingdom have come together to align approaches to agentic evaluations. This is the third exercise, building on insights from two earlier joint testing exercises conducted by the Network in November 2024 and February 2025. The objective is to further refine best practices for testing advanced AI systems. The exercise was split into two strands: (1) common risks, including leakage of sensitive information and fraud, led by Singapore AISI; and (2) cybersecurity, led by UK AISI. A mix of open and closed-weight models were evaluated against tasks from various public agentic benchmarks. Given the nascency of agentic testing, our primary focus was on understanding methodological issues in conducting such tests, rather than examining test results or model capabilities. This collaboration marks an important step forward as participants work together to advance the science of agentic evaluations.

Bushra Sabir, Shigang Liu, Seung Ick Jang et al.

Automatically generating source code from natural language using large language models (LLMs) is becoming common, yet security vulnerabilities persist despite advances in fine tuning and prompting. In this work, we systematically evaluate whether multi LLM ensembles and collaborative strategies can meaningfully improve secure code generation. We present MULTI-LLMSECCODEEVAL, a framework for assessing and enhancing security across the vulnerability management lifecycle by combining multiple LLMs with static analysis and structured collaboration. Using SecLLMEval and SecLLMHolmes, we benchmark ten pipelines spanning single model, ensemble, collaborative, and hybrid designs. Our results show that ensemble pipelines augmented with static analysis improve secure code generation over single LLM baselines by up to 47.3% on SecLLMEval and 19.3% on SecLLMHolmes, while purely LLM based collaborative pipelines yield smaller gains of 8.9% to 22.3%. Hybrid pipelines that integrate ensembling, detection, and patching achieve the strongest security performance, outperforming the best ensemble baseline by 1.78% to 4.72% and collaborative baselines by 19.81% to 26.78%. Ablation studies reveal that model scale alone does not ensure security. Smaller, structured multi model ensembles consistently outperform large monolithic LLMs. Overall, our findings demonstrate that secure code does not emerge from scale, but from carefully orchestrated multi model system design.

Thusitha Dayaratne, Ngoc Duy Pham, Viet Vo et al.

The quality and experience of mobile communication have significantly improved with the introduction of 5G, and these improvements are expected to continue beyond the 5G era. However, vulnerabilities in control-plane protocols, such as Radio Resource Control (RRC) and Non-Access Stratum (NAS), pose significant security threats, such as Blind Denial of Service (DoS) attacks. Despite the availability of existing anomaly detection methods that leverage rule-based systems or traditional machine learning methods, these methods have several limitations, including the need for extensive training data, predefined rules, and limited explainability. Addressing these challenges, we propose a novel anomaly detection framework that leverages the capabilities of Large Language Models (LLMs) in zero-shot mode with unordered data and short natural language attack descriptions within the Open Radio Access Network (O-RAN) architecture. We analyse robustness to prompt variation, demonstrate the practicality of automating the attack descriptions and show that detection quality relies on the semantic completeness of the description rather than its phrasing or length. We utilise an RRC/NAS dataset to evaluate the solution and provide an extensive comparison of open-source and proprietary LLM implementations to demonstrate superior performance in attack detection. We further validate the practicality of our framework within O-RAN's real-time constraints, illustrating its potential for detecting other Layer-3 attacks.

Thusitha Dayaratne, Ngoc Duy Pham, Viet Vo et al.

The introduction of 5G and the Open Radio Access Network (O-RAN) architecture has enabled more flexible and intelligent network deployments. However, the increased complexity and openness of these architectures also introduce novel security challenges, such as data manipulation attacks on the semi-standardised Shared Data Layer (SDL) within the O-RAN platform through malicious xApps. In particular, malicious xApps can exploit this vulnerability by introducing subtle Unicode-wise alterations (hypoglyphs) into the data that are being used by traditional machine learning (ML)-based anomaly detection methods. These Unicode-wise manipulations can potentially bypass detection and cause failures in anomaly detection systems based on traditional ML, such as AutoEncoders, which are unable to process hypoglyphed data without crashing. We investigate the use of Large Language Models (LLMs) for anomaly detection within the O-RAN architecture to address this challenge. We demonstrate that LLM-based xApps maintain robust operational performance and are capable of processing manipulated messages without crashing. While initial detection accuracy requires further improvements, our results highlight the robustness of LLMs to adversarial attacks such as hypoglyphs in input data. There is potential to use their adaptability through prompt engineering to further improve the accuracy, although this requires further research. Additionally, we show that LLMs achieve low detection latency (under 0.07 seconds), making them suitable for Near-Real-Time (Near-RT) RIC deployments.

Farina Riaz, Fakhar Zaman, Hajime Suzuki et al.

Variational autoencoders (VAEs) are fundamental for generative modeling and image reconstruction, yet their performance often struggles to maintain high fidelity in reconstructions. This study introduces a hybrid model, quantum variational autoencoder (Q-VAE), which integrates quantum encoding within the encoder while utilizing fully connected layers to extract meaningful representations. The decoder uses transposed convolution layers for up-sampling. The Q-VAE is evaluated against the classical VAE and the classical direct-passing VAE, which utilizes windowed pooling filters. Results on the MNIST and USPS datasets demonstrate that Q-VAE consistently outperforms classical approaches, achieving lower Fréchet inception distance scores, thereby indicating superior image fidelity and enhanced reconstruction quality. These findings highlight the potential of Q-VAE for high-quality synthetic data generation and improved image reconstruction in generative models.

Nazatul Haque Sultan, Yan Bo, Yansong Gao et al.

Federated Learning (FL) enables multiple users to collaboratively train a machine learning model without sharing raw data, making it suitable for privacy-sensitive applications. However, local model or weight updates can still leak sensitive information. Secure aggregation protocols mitigate this risk by ensuring that only the aggregated updates are revealed. Among these, single-setup protocols, where key generation and exchange occur only once, are the most efficient due to reduced communication and computation overhead. However, existing single-setup protocols often lack support for dynamic user participation and do not provide strong privacy guarantees such as forward and backward secrecy. \par In this paper, we present a novel secure aggregation protocol that requires only a single setup for the entire FL training. Our protocol supports dynamic user participation, tolerates dropouts, and achieves both forward and backward secrecy. It leverages lightweight symmetric homomorphic encryption with a key negation technique to mask updates efficiently, eliminating the need for user-to-user communication. To defend against model inconsistency attacks, we introduce a low-overhead verification mechanism using message authentication codes (MACs). We provide formal security proofs under both semi-honest and malicious adversarial models and implement a full prototype. Experimental results show that our protocol reduces user-side computation by up to $99\%$ compared to state-of-the-art protocols like e-SeaFL (ACSAC'24), while maintaining competitive model accuracy. These features make our protocol highly practical for real-world FL deployments, especially on resource-constrained devices.

David D. Nguyen, David Liebowitz, Surya Nepal et al.

Honeyfiles are security assets designed to attract and detect intruders on compromised systems. Honeyfiles are a type of honeypot that mimic real, sensitive documents, creating the illusion of the presence of valuable data. Interaction with a honeyfile reveals the presence of an intruder, and can provide insights into their goals and intentions. Their practical use, however, is limited by the time, cost and effort associated with manually creating realistic content. The introduction of large language models has made high-quality text generation accessible, but honeyfiles contain a variety of content including charts, tables and images. This content needs to be plausible and realistic, as well as semantically consistent both within honeyfiles and with the real documents they mimic, to successfully deceive an intruder. In this paper, we focus on an important component of the honeyfile content generation problem: document charts. Charts are ubiquitous in corporate documents and are commonly used to communicate quantitative and scientific data. Existing image generation models, such as DALL-E, are rather prone to generating charts with incomprehensible text and unconvincing data. We take a multi-modal approach to this problem by combining two purpose-built generative models: a multitask Transformer and a specialized multi-head autoencoder. The Transformer generates realistic captions and plot text, while the autoencoder generates the underlying tabular data for the plot. To advance the field of automated honeyplot generation, we also release a new document-chart dataset and propose a novel metric Keyword Semantic Matching (KSM). This metric measures the semantic consistency between keywords of a corpus and a smaller bag of words. Extensive experiments demonstrate excellent performance against multiple large language models, including ChatGPT and GPT4.

Ngoc Duy Pham, Thusitha Dayaratne, Viet Vo et al.

Advancements in wireless and mobile technologies, including 5G advanced and the envisioned 6G, are driving exponential growth in wireless devices. However, this rapid expansion exacerbates spectrum scarcity, posing a critical challenge. Dynamic spectrum allocation (DSA)--which relies on sensing and dynamically sharing spectrum--has emerged as an essential solution to address this issue. While machine learning (ML) models hold significant potential for improving spectrum sensing, their adoption in centralized ML-based DSA systems is limited by privacy concerns, bandwidth constraints, and regulatory challenges. To overcome these limitations, distributed ML-based approaches such as Federated Learning (FL) offer promising alternatives. This work addresses two key challenges in FL-based spectrum sensing (FLSS). First, the scarcity of labeled data for training FL models in practical spectrum sensing scenarios is tackled with a semi-supervised FL approach, combined with energy detection, enabling model training on unlabeled datasets. Second, we examine the security vulnerabilities of FLSS, focusing on the impact of data poisoning attacks. Our analysis highlights the shortcomings of existing majority-based defenses in countering such attacks. To address these vulnerabilities, we propose a novel defense mechanism inspired by vaccination, which effectively mitigates data poisoning attacks without relying on majority-based assumptions. Extensive experiments on both synthetic and real-world datasets validate our solutions, demonstrating that FLSS can achieve near-perfect accuracy on unlabeled datasets and maintain Byzantine robustness against both targeted and untargeted data poisoning attacks, even when a significant proportion of participants are malicious.

Viet Vo, Thusitha Dayaratne, Blake Haydon et al.

Spectrum sharing is increasingly vital in 6G wireless communication, facilitating dynamic access to unused spectrum holes. Recently, there has been a significant shift towards employing machine learning (ML) techniques for sensing spectrum holes. In this context, federated learning (FL)-enabled spectrum sensing technology has garnered wide attention, allowing for the construction of an aggregated ML model without disclosing the private spectrum sensing information of wireless user devices. However, the integrity of collaborative training and the privacy of spectrum information from local users have remained largely unexplored. This article first examines the latest developments in FL-enabled spectrum sharing for prospective 6G scenarios. It then identifies practical attack vectors in 6G to illustrate potential AI-powered security and privacy threats in these contexts. Finally, the study outlines future directions, including practical defense challenges and guidelines.

Mahathir Almashor, Ejaz Ahmed, Benjamin Pick et al.

URLs are central to a myriad of cyber-security threats, from phishing to the distribution of malware. Their inherent ease of use and familiarity is continuously abused by attackers to evade defences and deceive end-users. Seemingly dissimilar URLs are being used in an organized way to perform phishing attacks and distribute malware. We refer to such behaviours as campaigns, with the hypothesis being that attacks are often coordinated to maximize success rates and develop evasion tactics. The aim is to gain better insights into campaigns, bolster our grasp of their characteristics, and thus aid the community devise more robust solutions. To this end, we performed extensive research and analysis into 311M records containing 77M unique real-world URLs that were submitted to VirusTotal from Dec 2019 to Jan 2020. From this dataset, 2.6M suspicious campaigns were identified based on their attached metadata, of which 77,810 were doubly verified as malicious. Using the 38.1M records and 9.9M URLs within these malicious campaigns, we provide varied insights such as their targeted victim brands as well as URL sizes and heterogeneity. Some surprising findings were observed, such as detection rates falling to just 13.27% for campaigns that employ more than 100 unique URLs. The paper concludes with several case-studies that illustrate the common malicious techniques employed by attackers to imperil users and circumvent defences.

Jihyeon Ryu, Yifeng Zheng, Yansong Gao et al.

Collaborative inference has recently emerged as an attractive framework for applying deep learning to Internet of Things (IoT) applications by splitting a DNN model into several subpart models among resource-constrained IoT devices and the cloud. However, the reconstruction attack was proposed recently to recover the original input image from intermediate outputs that can be collected from local models in collaborative inference. For addressing such privacy issues, a promising technique is to adopt differential privacy so that the intermediate outputs are protected with a small accuracy loss. In this paper, we provide the first systematic study to reveal insights regarding the effectiveness of differential privacy for collaborative inference against the reconstruction attack. We specifically explore the privacy-accuracy trade-offs for three collaborative inference models with four datasets (SVHN, GTSRB, STL-10, and CIFAR-10). Our experimental analysis demonstrates that differential privacy can practically be applied to collaborative inference when a dataset has small intra-class variations in appearance. With the (empirically) optimized privacy budget parameter in our study, the differential privacy technique incurs accuracy loss of 0.476%, 2.066%, 5.021%, and 12.454% on SVHN, GTSRB, STL-10, and CIFAR-10 datasets, respectively, while thwarting the reconstruction attack.

Yansong Gao, Minki Kim, Chandra Thapa et al.

Federated learning (FL) and split learning (SL) are state-of-the-art distributed machine learning techniques to enable machine learning training without accessing raw data on clients or end devices. However, their \emph{comparative training performance} under real-world resource-restricted Internet of Things (IoT) device settings, e.g., Raspberry Pi, remains barely studied, which, to our knowledge, have not yet been evaluated and compared, rendering inconvenient reference for practitioners. This work firstly provides empirical comparisons of FL and SL in real-world IoT settings regarding (i) learning performance with heterogeneous data distributions and (ii) on-device execution overhead. Our analyses in this work demonstrate that the learning performance of SL is better than FL under an imbalanced data distribution but worse than FL under an extreme non-IID data distribution. Recently, FL and SL are combined to form splitfed learning (SFL) to leverage each of their benefits (e.g., parallel training of FL and lightweight on-device computation requirement of SL). This work then considers FL, SL, and SFL, and mount them on Raspberry Pi devices to evaluate their performance, including training time, communication overhead, power consumption, and memory usage. Besides evaluations, we apply two optimizations. Firstly, we generalize SFL by carefully examining the possibility of a hybrid type of model training at the server-side. The generalized SFL merges sequential (dependent) and parallel (independent) processes of model training and is thus beneficial for a system with large-scaled IoT devices, specifically at the server-side operations. Secondly, we propose pragmatic techniques to substantially reduce the communication overhead by up to four times for the SL and (generalized) SFL.

Bedeuro Kim, Sharif Abuadbba, Hyoungshick Kim

Image spam emails are often used to evade text-based spam filters that detect spam emails with their frequently used keywords. In this paper, we propose a new image spam email detection tool called DeepCapture using a convolutional neural network (CNN) model. There have been many efforts to detect image spam emails, but there is a significant performance degrade against entirely new and unseen image spam emails due to overfitting during the training phase. To address this challenging issue, we mainly focus on developing a more robust model to address the overfitting problem. Our key idea is to build a CNN-XGBoost framework consisting of eight layers only with a large number of training samples using data augmentation techniques tailored towards the image spam detection task. To show the feasibility of DeepCapture, we evaluate its performance with publicly available datasets consisting of 6,000 spam and 2,313 non-spam image samples. The experimental results show that DeepCapture is capable of achieving an F1-score of 88%, which has a 6% improvement over the best existing spam detection model CNN-SVM with an F1-score of 82%. Moreover, DeepCapture outperformed existing image spam detection solutions against new and unseen image datasets.

Yansong Gao, Minki Kim, Sharif Abuadbba et al.

This work is the first attempt to evaluate and compare felderated learning (FL) and split neural networks (SplitNN) in real-world IoT settings in terms of learning performance and device implementation overhead. We consider a variety of datasets, different model architectures, multiple clients, and various performance metrics. For learning performance, which is specified by the model accuracy and convergence speed metrics, we empirically evaluate both FL and SplitNN under different types of data distributions such as imbalanced and non-independent and identically distributed (non-IID) data. We show that the learning performance of SplitNN is better than FL under an imbalanced data distribution, but worse than FL under an extreme non-IID data distribution. For implementation overhead, we end-to-end mount both FL and SplitNN on Raspberry Pis, and comprehensively evaluate overheads including training time, communication overhead under the real LAN setting, power consumption and memory usage. Our key observations are that under IoT scenario where the communication traffic is the main concern, the FL appears to perform better over SplitNN because FL has the significantly lower communication overhead compared with SplitNN, which empirically corroborate previous statistical analysis. In addition, we reveal several unrecognized limitations about SplitNN, forming the basis for future research.

Sharif Abuadbba, Kyuyeon Kim, Minki Kim et al.

A new collaborative learning, called split learning, was recently introduced, aiming to protect user data privacy without revealing raw input data to a server. It collaboratively runs a deep neural network model where the model is split into two parts, one for the client and the other for the server. Therefore, the server has no direct access to raw data processed at the client. Until now, the split learning is believed to be a promising approach to protect the client's raw data; for example, the client's data was protected in healthcare image applications using 2D convolutional neural network (CNN) models. However, it is still unclear whether the split learning can be applied to other deep learning models, in particular, 1D CNN. In this paper, we examine whether split learning can be used to perform privacy-preserving training for 1D CNN models. To answer this, we first design and implement an 1D CNN model under split learning and validate its efficacy in detecting heart abnormalities using medical ECG data. We observed that the 1D CNN model under split learning can achieve the same accuracy of 98.9\% like the original (non-split) model. However, our evaluation demonstrates that split learning may fail to protect the raw data privacy on 1D CNN models. To address the observed privacy leakage in split learning, we adopt two privacy leakage mitigation techniques: 1) adding more hidden layers to the client side and 2) applying differential privacy. Although those mitigation techniques are helpful in reducing privacy leakage, they have a significant impact on model accuracy. Hence, based on those results, we conclude that split learning alone would not be sufficient to maintain the confidentiality of raw sequential data in 1D CNN models.

Sharif Abuadbba, Ayman Ibaida, Ibrahim Khalil

Remotely generated data by Intent of Things (IoT) has recently had a lot of attention for their huge benefits such as efficient monitoring and risk reduction. The transmitted streams usually consist of periodical streams (e.g. activities) and highly private information (e.g. IDs). Despite the obvious benefits, the concerns are the secrecy and the originality of the transferred data. Surprisingly, although these concerns have been well studied for static data, they have received only limited attention for streaming data. Therefore, this paper introduces a new steganographic mechanism that provides (1) robust privacy protection of secret information by concealing them arbitrarily in the transported readings employing a random key, and (2) permanent proof of originality for the normal streams. This model surpasses our previous works by employing the Discrete Cosine Transform to expand the hiding capacity and reduce complexity. The resultant distortion has been accurately measured at all stages - the original, the stego, and the recovered forms - using a well-known measurement matrix called Percentage Residual Difference (PRD). After thorough experiments on three types of streams (i.e. chemical, environmental and smart homes), it has been proven that the original streams have not been affected (< 1 %). Also, the mathematical analysis shows that the model has much lighter (i.e. linear) computational complexity O(n) compared to existing work.