Search

Speaker identification (SI) determines a speaker's identity based on their spoken utterances. Previous work indicates that SI deep neural networks (DNNs) are vulnerable to backdoor attacks. Backdoor attacks involve embedding hidden triggers in DNNs' training data, causing the DNN to produce incorrect output when these triggers are present during inference. This is the first work that explores SI DNNs' vulnerability to backdoor attacks using speakers' emotional prosody, resulting in dynamic, inconspicuous triggers. %Such an attack could have real-world implications in forensics, authentication, and surveillance. We conducted a parameter study using three different datasets and DNN architectures to determine the impact of emotions as backdoor triggers on the accuracy of SI systems. Additionally, we have explored the robustness of our attacks by applying defenses like pruning, STRIP-ViTA, and three popular preprocessing techniques: quantization, median filtering, and squeezing. Our findings show that the aforementioned models are prone to our attack, indicating that emotional triggers (sad and neutral prosody) can be effectively used to compromise the integrity of SI systems. However, the results of our pruning experiments suggest potential solutions for reinforcing the models against our attacks, decreasing the attack success rate up to 40%.

In this paper, we introduce MoRSE (Mixture of RAGs Security Experts), the first specialised AI chatbot for cybersecurity. MoRSE aims to provide comprehensive and complete knowledge about cybersecurity. MoRSE uses two RAG (Retrieval Augmented Generation) systems designed to retrieve and organize information from multidimensional cybersecurity contexts. MoRSE differs from traditional RAGs by using parallel retrievers that work together to retrieve semantically related information in different formats and structures. Unlike traditional Large Language Models (LLMs) that rely on Parametric Knowledge Bases, MoRSE retrieves relevant documents from Non-Parametric Knowledge Bases in response to user queries. Subsequently, MoRSE uses this information to generate accurate answers. In addition, MoRSE benefits from real-time updates to its knowledge bases, enabling continuous knowledge enrichment without retraining. We have evaluated the effectiveness of MoRSE against other state-of-the-art LLMs, evaluating the system on 600 cybersecurity specific questions. The experimental evaluation has shown that the improvement in terms of relevance and correctness of the answer is more than 10\% compared to known solutions such as GPT-4 and Mixtral 7x8.

The Remote ID (RID) regulation recently introduced by several aviation authorities worldwide (including the US and EU) forces commercial drones to regularly (max. every second) broadcast plaintext messages on the wireless channel, providing information about the drone identifier and current location, among others. Although these regulations increase the accountability of drone operations and improve traffic management, they allow malicious users to track drones via the disclosed information, possibly leading to drone capture and severe privacy leaks. In this paper, we propose Obfuscated Location disclOsure for RID-enabled drones (OLO-RID), a solution modifying and extending the RID regulation while preserving drones' location privacy. Rather than disclosing the actual drone's location, drones equipped with OLO-RID disclose a differentially private obfuscated location in a mobile scenario. OLO-RID also extends RID messages with encrypted location information, accessible only by authorized entities and valuable to obtain the current drone's location in safety-critical use cases. We design, implement, and deploy OLO-RID on a Raspberry Pi 3 and release the code of our implementation as open-source. We also perform an extensive performance assessment of the runtime overhead of our solution in terms of processing, communication, memory, and energy consumption. We show that OLO-RID can generate RID messages on a constrained device in less than 0.16 s while also requiring a minimal energy toll on a relevant device (0.0236% of energy for a DJI Mini 2). We also evaluate the utility of the proposed approach in the context of three reference use cases involving the drones' location usage, demonstrating minimal performance degradation when trading off location privacy and utility for next-generation RID-compliant drone ecosystems., Comment: 18 pages, 8 figures. Submitted and under review for journal publication

Despite its well-known security issues, the Controller Area Network (CAN) is still the main technology for in-vehicle communications. Attackers posing as diagnostic services or accessing the CAN bus can threaten the drivers' location privacy to know the exact location at a certain point in time or to infer the visited areas. This represents a serious threat to users' privacy, but also an advantage for police investigations to gather location-based evidence. In this paper, we present On Path Diagnostic - Intrusion \& Inference (OPD-II), a novel path inference attack leveraging a physical car model and a map matching algorithm to infer the path driven by a car based on CAN bus data. Differently from available attacks, our approach only requires the attacker to know the initial location and heading of the victim's car and is not limited by the availability of training data, road configurations, or the need to access other victim's devices (e.g., smartphones). We implement our attack on a set of four different cars and a total number of 41 tracks in different road and traffic scenarios. We achieve an average of 95% accuracy on reconstructing the coordinates of the recorded path by leveraging a dynamic map-matching algorithm that outperforms the 75% and 89% accuracy values of other proposals while removing their set of assumptions.

Generative models are gaining significant attention as potential catalysts for a novel industrial revolution. Since automated sample generation can be useful to solve privacy and data scarcity issues that usually affect learned biometric models, such technologies became widely spread in this field. In this paper, we assess the vulnerabilities of generative machine learning models concerning identity protection by designing and testing an identity inference attack on fingerprint datasets created by means of a generative adversarial network. Experimental results show that the proposed solution proves to be effective under different configurations and easily extendable to other biometric measurements., Comment: Paper submitted at "Pattern Recognition Letters", 9 pages, 6 images

The rise of IT-dependent operations in modern organizations has heightened their vulnerability to cyberattacks. As a growing number of organizations include smart, interconnected devices in their systems to automate their processes, the attack surface becomes much bigger, and the complexity and frequency of attacks pose a significant threat. Consequently, organizations have been compelled to seek innovative approaches to mitigate the menaces inherent in their infrastructure. In response, considerable research efforts have been directed towards creating effective solutions for sharing Cyber Threat Intelligence (CTI). Current information-sharing methods lack privacy safeguards, leaving organizations vulnerable to leaks of both proprietary and confidential data. To tackle this problem, we designed a novel framework called SeCTIS (Secure Cyber Threat Intelligence Sharing), integrating Swarm Learning and Blockchain technologies to enable businesses to collaborate, preserving the privacy of their CTI data. Moreover, our approach provides a way to assess the data and model quality, and the trustworthiness of all the participants leveraging some validators through Zero Knowledge Proofs. An extensive experimental campaign demonstrates our framework's correctness and performance, and the detailed attack model discusses its robustness against attacks in the context of data and model quality.

Malware has become a formidable threat as it has been growing exponentially in number and sophistication, thus, it is imperative to have a solution that is easy to implement, reliable, and effective. While recent research has introduced deep learning multi-feature fusion algorithms, they lack a proper explanation. In this work, we investigate the power of fusing Convolutional Neural Network models trained on different modalities of a malware executable. We are proposing a novel multimodal fusion algorithm, leveraging three different visual malware features: Grayscale Image, Entropy Graph, and SimHash Image, with which we conducted exhaustive experiments independently on each feature and combinations of all three of them using fusion operators such as average, maximum, add, and concatenate for effective malware detection and classification. The proposed strategy has a detection rate of 1.00 (on a scale of 0-1) in identifying malware in the given dataset. We explained its interpretability with visualization techniques such as t-SNE and Grad-CAM. Experimental results show the model works even for a highly imbalanced dataset. We also assessed the effectiveness of the proposed method on obfuscated malware and achieved state-of-the-art results. The proposed methodology is more reliable as our findings prove VGG16 model can detect and classify malware in a matter of seconds in real-time.

The smart grid represents a pivotal innovation in modernizing the electricity sector, offering an intelligent, digitalized energy network capable of optimizing energy delivery from source to consumer. It hence represents the backbone of the energy sector of a nation. Due to its central role, the availability of the smart grid is paramount and is hence necessary to have in-depth control of its operations and safety. To this aim, researchers developed multiple solutions to assess the smart grid's stability and guarantee that it operates in a safe state. Artificial intelligence and Machine learning algorithms have proven to be effective measures to accurately predict the smart grid's stability. Despite the presence of known adversarial attacks and potential solutions, currently, there exists no standardized measure to protect smart grids against this threat, leaving them open to new adversarial attacks. In this paper, we propose GAN-GRID a novel adversarial attack targeting the stability prediction system of a smart grid tailored to real-world constraints. Our findings reveal that an adversary armed solely with the stability model's output, devoid of data or model knowledge, can craft data classified as stable with an Attack Success Rate (ASR) of 0.99. Also by manipulating authentic data and sensor values, the attacker can amplify grid issues, potentially undetected due to a compromised stability prediction system. These results underscore the imperative of fortifying smart grid security mechanisms against adversarial manipulation to uphold system stability and reliability.

Autonomous driving is a research direction that has gained enormous traction in the last few years thanks to advancements in Artificial Intelligence (AI). Depending on the level of independence from the human driver, several studies show that Autonomous Vehicles (AVs) can reduce the number of on-road crashes and decrease overall fuel emissions by improving efficiency. However, security research on this topic is mixed and presents some gaps. On one hand, these studies often neglect the intrinsic vulnerabilities of AI algorithms, which are known to compromise the security of these systems. On the other, the most prevalent attacks towards AI rely on unrealistic assumptions, such as access to the model parameters or the training dataset. As such, it is unclear if autonomous driving can still claim several advantages over human driving in real-world applications. This paper evaluates the inherent risks in autonomous driving by examining the current landscape of AVs and establishing a pragmatic threat model. Through our analysis, we develop specific claims highlighting the delicate balance between the advantages of AVs and potential security challenges in real-world scenarios. Our evaluation serves as a foundation for providing essential takeaway messages, guiding both researchers and practitioners at various stages of the automation pipeline. In doing so, we contribute valuable insights to advance the discourse on the security and viability of autonomous driving in real-world applications., Comment: Accepted at ACSW 2024

In this paper, we investigate the security implications of virtualized and software-based Open Radio Access Network (RAN) systems, specifically focusing on the architecture proposed by the O-RAN ALLIANCE and O-Cloud deployments based on the O-RAN Software Community (OSC) stack and infrastructure. Our key findings are based on a thorough security assessment and static scanning of the OSC Near Real-Time RAN Intelligent Controller (RIC) cluster. We highlight the presence of potential vulnerabilities and misconfigurations in the Kubernetes infrastructure supporting the RIC, also due to the usage of outdated versions of software packages, and provide an estimation of their criticality using various deployment auditing frameworks (e.g., MITRE ATT&CK and the NSA CISA). In addition, we propose methodologies to minimize these issues and harden the Open RAN virtualization infrastructure. These encompass the integration of security evaluation methods into the deployment process, implementing deployment hardening measures, and employing policy-based control for RAN components. We emphasize the need to address the problems found in order to improve the overall security of virtualized Open RAN systems.

Web Vulnerability Assessment and Penetration Testing (Web VAPT) is a comprehensive cybersecurity process that uncovers a range of vulnerabilities which, if exploited, could compromise the integrity of web applications. In a VAPT, it is common to perform a \textit{Directory brute-forcing Attack}, aiming at the identification of accessible directories of a target website. Current commercial solutions are inefficient as they are based on brute-forcing strategies that use wordlists, resulting in enormous quantities of trials for a small amount of success. Offensive AI is a recent paradigm that integrates AI-based technologies in cyber attacks. In this work, we explore whether AI can enhance the directory enumeration process and propose a novel Language Model-based framework. Our experiments -- conducted in a testbed consisting of 1 million URLs from different web application domains (universities, hospitals, government, companies) -- demonstrate the superiority of the LM-based attack, with an average performance increase of 969%., Comment: Under submission

Recent advancements in Artificial Intelligence, and particularly Large Language Models (LLMs), offer promising prospects for aiding system administrators in managing the complexity of modern networks. However, despite this potential, a significant gap exists in the literature regarding the extent to which LLMs can understand computer networks. Without empirical evidence, system administrators might rely on these models without assurance of their efficacy in performing network-related tasks accurately. In this paper, we are the first to conduct an exhaustive study on LLMs' comprehension of computer networks. We formulate several research questions to determine whether LLMs can provide correct answers when supplied with a network topology and questions on it. To assess them, we developed a thorough framework for evaluating LLMs' capabilities in various network-related tasks. We evaluate our framework on multiple computer networks employing proprietary (e.g., GPT4) and open-source (e.g., Llama2) models. Our findings in general purpose LLMs using a zero-shot scenario demonstrate promising results, with the best model achieving an average accuracy of 79.3%. Proprietary LLMs achieve noteworthy results in small and medium networks, while challenges persist in comprehending complex network topologies, particularly for open-source models. Moreover, we provide insight into how prompt engineering can enhance the accuracy of some tasks.

The widespread exchange of digital documents in various domains has resulted in abundant private information being shared. This proliferation necessitates redaction techniques to protect sensitive content and user privacy. While numerous redaction methods exist, their effectiveness varies, with some proving more robust than others. As such, the literature proposes several deanonymization techniques, raising awareness of potential privacy threats. However, while none of these methods are successful against the most effective redaction techniques, these attacks only focus on the anonymized tokens and ignore the sentence context. In this paper, we propose RedactBuster, the first deanonymization model using sentence context to perform Named Entity Recognition on reacted text. Our methodology leverages fine-tuned state-of-the-art Transformers and Deep Learning models to determine the anonymized entity types in a document. We test RedactBuster against the most effective redaction technique and evaluate it using the publicly available Text Anonymization Benchmark (TAB). Our results show accuracy values up to 0.985 regardless of the document nature or entity type. In raising awareness of this privacy issue, we propose a countermeasure we call character evasion that helps strengthen the secrecy of sensitive information. Furthermore, we make our model and testbed open-source to aid researchers and practitioners in evaluating the resilience of novel redaction techniques and enhancing document privacy.

The growing integration of vehicles with external networks has led to a surge in attacks targeting their Controller Area Network (CAN) internal bus. As a countermeasure, various Intrusion Detection Systems (IDSs) have been suggested in the literature to prevent and mitigate these threats. With the increasing volume of data facilitated by the integration of Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communication networks, most of these systems rely on data-driven approaches such as Machine Learning (ML) and Deep Learning (DL) models. However, these systems are susceptible to adversarial evasion attacks. While many researchers have explored this vulnerability, their studies often involve unrealistic assumptions, lack consideration for a realistic threat model, and fail to provide effective solutions. In this paper, we present CANEDERLI (CAN Evasion Detection ResiLIence), a novel framework for securing CAN-based IDSs. Our system considers a realistic threat model and addresses the impact of adversarial attacks on DL-based detection systems. Our findings highlight strong transferability properties among diverse attack methodologies by considering multiple state-of-the-art attacks and model architectures. We analyze the impact of adversarial training in addressing this threat and propose an adaptive online adversarial training technique outclassing traditional fine-tuning methodologies with F1 scores up to 0.941. By making our framework publicly available, we aid practitioners and researchers in assessing the resilience of IDSs to a varied adversarial landscape., Comment: Accepted at WiseML 2024

Machine learning based phishing website detectors (ML-PWD) are a critical part of today's anti-phishing solutions in operation. Unfortunately, ML-PWD are prone to adversarial evasions, evidenced by both academic studies and analyses of real-world adversarial phishing webpages. However, existing works mostly focused on assessing adversarial phishing webpages against ML-PWD, while neglecting a crucial aspect: investigating whether they can deceive the actual target of phishing -- the end users. In this paper, we fill this gap by conducting two user studies (n=470) to examine how human users perceive adversarial phishing webpages, spanning both synthetically crafted ones (which we create by evading a state-of-the-art ML-PWD) as well as real adversarial webpages (taken from the wild Web) that bypassed a production-grade ML-PWD. Our findings confirm that adversarial phishing is a threat to both users and ML-PWD, since most adversarial phishing webpages have comparable effectiveness on users w.r.t. unperturbed ones. However, not all adversarial perturbations are equally effective. For example, those with added typos are significantly more noticeable to users, who tend to overlook perturbations of higher visual magnitude (such as replacing the background). We also show that users' self-reported frequency of visiting a brand's website has a statistically negative correlation with their phishing detection accuracy, which is likely caused by overconfidence. We release our resources.

Predicting and classifying faults in electricity networks is crucial for uninterrupted provision and keeping maintenance costs at a minimum. Thanks to the advancements in the field provided by the smart grid, several data-driven approaches have been proposed in the literature to tackle fault prediction tasks. Implementing these systems brought several improvements, such as optimal energy consumption and quick restoration. Thus, they have become an essential component of the smart grid. However, the robustness and security of these systems against adversarial attacks have not yet been extensively investigated. These attacks can impair the whole grid and cause additional damage to the infrastructure, deceiving fault detection systems and disrupting restoration. In this paper, we present FaultGuard, the first framework for fault type and zone classification resilient to adversarial attacks. To ensure the security of our system, we employ an Anomaly Detection System (ADS) leveraging a novel Generative Adversarial Network training layer to identify attacks. Furthermore, we propose a low-complexity fault prediction model and an online adversarial training technique to enhance robustness. We comprehensively evaluate the framework's performance against various adversarial attacks using the IEEE13-AdvAttack dataset, which constitutes the state-of-the-art for resilient fault prediction benchmarking. Our model outclasses the state-of-the-art even without considering adversaries, with an accuracy of up to 0.958. Furthermore, our ADS shows attack detection capabilities with an accuracy of up to 1.000. Finally, we demonstrate how our novel training layers drastically increase performances across the whole framework, with a mean increase of 154% in ADS accuracy and 118% in model accuracy.

Sharding enhances blockchain scalability by dividing the network into shards, each managing specific unspent transaction outputs or accounts. As an introduced new transaction type, cross-shard transactions pose a critical challenge to the security and efficiency of sharding blockchains. Currently, there is a lack of a generic sharding consensus pattern that achieves both security and low overhead. In this paper, we present Kronos, a secure sharding blockchain consensus achieving optimized overhead. In particular, we propose a new secure sharding consensus pattern, based on a buffer managed jointly by shard members. Valid transactions are transferred to the payee via the buffer, while invalid ones are rejected through happy or unhappy paths. Kronos is proved to achieve security with atomicity under malicious clients with optimal intra-shard overhead $kB$ ($k$ for involved shard number and $B$ for a Byzantine fault tolerance (BFT) cost). Besides, we propose secure cross-shard certification methods based on batch certification and reliable cross-shard transfer. The former combines hybrid trees or vector commitments, while the latter integrates erasure coding. Handling $b$ transactions, Kronos is proved to achieve reliability with low cross-shard overhead $O(n b \lambda)$ ($n$ for shard size and $\lambda$ for the security parameter). Notably, Kronos imposes no restrictions on BFT and does not rely on time assumptions, offering optional constructions in various modules. We implement Kronos using two prominent BFT protocols: asynchronous Speeding Dumbo and partial synchronous Hotstuff. Extensive experiments demonstrate Kronos scales the consensus nodes to thousands, achieving a substantial throughput of 320 ktx/sec with 2.0 sec latency. Compared with the past solutions, Kronos outperforms, achieving up to a 12* improvement in throughput and a 50% reduction in latency.

Recent advancements in drone technology have shown that commercial off-the-shelf Micro Aerial Drones are more effective than large-sized drones for performing flight missions in narrow environments, such as swarming, indoor navigation, and inspection of hazardous locations. Due to their deployments in many civilian and military applications, safe and reliable communication of these drones throughout the mission is critical. The Crazyflie ecosystem is one of the most popular Micro Aerial Drones and has the potential to be deployed worldwide. In this paper, we empirically investigate two interference attacks against the Crazy Real Time Protocol (CRTP) implemented within the Crazyflie drones. In particular, we explore the feasibility of experimenting two attack vectors that can disrupt an ongoing flight mission: the jamming attack, and the hijacking attack. Our experimental results demonstrate the effectiveness of such attacks in both autonomous and non-autonomous flight modes on a Crazyflie 2.1 drone. Finally, we suggest potential shielding strategies that guarantee a safe and secure flight mission. To the best of our knowledge, this is the first work investigating jamming and hijacking attacks against Micro Aerial Drones, both in autonomous and non-autonomous modes., Comment: Accepted at IEEE International Conference on Communications (ICC) 2024

Federated Learning (FL) is a machine learning (ML) approach that enables multiple decentralized devices or edge servers to collaboratively train a shared model without exchanging raw data. During the training and sharing of model updates between clients and servers, data and models are susceptible to different data-poisoning attacks. In this study, our motivation is to explore the severity of data poisoning attacks in the computer network domain because they are easy to implement but difficult to detect. We considered two types of data-poisoning attacks, label flipping (LF) and feature poisoning (FP), and applied them with a novel approach. In LF, we randomly flipped the labels of benign data and trained the model on the manipulated data. For FP, we randomly manipulated the highly contributing features determined using the Random Forest algorithm. The datasets used in this experiment were CIC and UNSW related to computer networks. We generated adversarial samples using the two attacks mentioned above, which were applied to a small percentage of datasets. Subsequently, we trained and tested the accuracy of the model on adversarial datasets. We recorded the results for both benign and manipulated datasets and observed significant differences between the accuracy of the models on different datasets. From the experimental results, it is evident that the LF attack failed, whereas the FP attack showed effective results, which proved its significance in fooling a server. With a 1% LF attack on the CIC, the accuracy was approximately 0.0428 and the ASR was 0.9564; hence, the attack is easily detectable, while with a 1% FP attack, the accuracy and ASR were both approximately 0.9600, hence, FP attacks are difficult to detect. We repeated the experiment with different poisoning percentages.

Thanks to technologies such as virtual network function the Fifth Generation (5G) of mobile networks dynamically allocate resources to different types of users in an on-demand fashion. Virtualization extends up to the 5G core, where software-defined networks and network slicing implement a customizable environment. These technologies can be controlled via application programming interfaces and web technologies, inheriting hence their security risks and settings. An attacker exploiting vulnerable implementations of the 5G core may gain privileged control of the network assets and disrupt its availability. However, there is currently no security assessment of the web security of the 5G core network. In this paper, we present the first security assessment of the 5G core from a web security perspective. We use the STRIDE threat modeling approach to define a complete list of possible threat vectors and associated attacks. Thanks to a suite of security testing tools, we cover all of these threats and test the security of the 5G core. In particular, we test the three most relevant open-source 5G core implementations, i.e., Open5GS, Free5Gc, and OpenAirInterface. Our analysis shows that all these cores are vulnerable to at least two of our identified attack vectors, demanding increased security measures in the development of future 5G core networks.

In this work, we present a device-centric analysis of security and privacy attacks and defenses on Extended Reality (XR) devices, highlighting the need for robust and privacy-aware security mechanisms. Based on our analysis, we present future research directions and propose design considerations to help ensure the security and privacy of XR devices., Comment: This is the author's version of the work. It is posted here for personal/educational use only. The definitive version was published in IEEE Security & Privacy Magazine Jan/Feb 2024

Digital identity has always been considered the keystone for implementing secure and trustworthy communications among parties. The ever-evolving digital landscape has gone through many technological transformations that have also affected the way entities are digitally identified. During this digital evolution, identity management has shifted from centralized to decentralized approaches. The last era of this journey is represented by the emerging Self-Sovereign Identity (SSI), which gives users full control over their data. SSI leverages decentralized identifiers (DIDs) and verifiable credentials (VCs), which have been recently standardized by the World Wide Web Community (W3C). These technologies have the potential to build more secure and decentralized digital identity systems, remarkably contributing to strengthening the security of communications that typically involve many distributed participants. It is worth noting that the scope of DIDs and VCs extends beyond individuals, encompassing a broad range of entities including cloud, edge, and Internet of Things (IoT) resources. However, due to their novelty, existing literature lacks a comprehensive survey on how DIDs and VCs have been employed in different application domains, which go beyond SSI systems. This paper provides readers with a comprehensive overview of such technologies from different perspectives. Specifically, we first provide the background on DIDs and VCs. Then, we analyze available implementations and offer an in-depth review of how these technologies have been employed across different use-case scenarios. Furthermore, we examine recent regulations and initiatives that have been emerging worldwide. Finally, we present some challenges that hinder their adoption in real-world scenarios and future research directions., Comment: 30 pages, 15 figures, and 9 tables

In the age of digital music streaming, playlists on platforms like Spotify have become an integral part of individuals' musical experiences. People create and publicly share their own playlists to express their musical tastes, promote the discovery of their favorite artists, and foster social connections. These publicly accessible playlists transcend the boundaries of mere musical preferences: they serve as sources of rich insights into users' attributes and identities. For example, the musical preferences of elderly individuals may lean more towards Frank Sinatra, while Billie Eilish remains a favored choice among teenagers. These playlists thus become windows into the diverse and evolving facets of one's musical identity. In this work, we investigate the relationship between Spotify users' attributes and their public playlists. In particular, we focus on identifying recurring musical characteristics associated with users' individual attributes, such as demographics, habits, or personality traits. To this end, we conducted an online survey involving 739 Spotify users, yielding a dataset of 10,286 publicly shared playlists encompassing over 200,000 unique songs and 55,000 artists. Through extensive statistical analyses, we first assess a deep connection between a user's Spotify playlists and their real-life attributes. For instance, we found individuals high in openness often create playlists featuring a diverse array of artists, while female users prefer Pop and K-pop music genres. Building upon these observed associations, we create accurate predictive models for users' attributes, presenting a novel DeepSet application that outperforms baselines in most of these users' attributes.

The large number of Electronic Control Units (ECUs) mounted on modern cars and their expansive communication capabilities create a substantial attack surface for potential exploitation. Despite the evolution of automotive technology, the continued use of the originally insecure Controller Area Network (CAN) bus leaves in-vehicle communications inherently non-secure. In response to the absence of standardized authentication protocols within the automotive domain, researchers propose diverse solutions, each with unique strengths and vulnerabilities. However, the continuous influx of new protocols and potential oversights in meeting security requirements and essential operational features further complicate the implementability of these protocols. This paper comprehensively reviews and compares the 15 most prominent authentication protocols for the CAN bus. Our analysis emphasizes their strengths and weaknesses, evaluating their alignment with critical security requirements for automotive authentication. Additionally, we evaluate protocols based on essential operational criteria that contribute to ease of implementation in predefined infrastructures, enhancing overall reliability and reducing the probability of successful attacks. Our study reveals a prevalent focus on defending against external attackers in existing protocols, exposing vulnerabilities to internal threats. Notably, authentication protocols employing hash chains, Mixed Message Authentication Codes, and asymmetric encryption techniques emerge as the most effective approaches. Through our comparative study, we classify the considered protocols based on their security attributes and suitability for implementation, providing valuable insights for future developments in the field., Comment: This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible

Federated Learning (FL) has recently arisen as a revolutionary approach to collaborative training Machine Learning models. According to this novel framework, multiple participants train a global model collaboratively, coordinating with a central aggregator without sharing their local data. As FL gains popularity in diverse domains, security, and privacy concerns arise due to the distributed nature of this solution. Therefore, integrating this strategy with Blockchain technology has been consolidated as a preferred choice to ensure the privacy and security of participants. This paper explores the research efforts carried out by the scientific community to define privacy solutions in scenarios adopting Blockchain-Enabled FL. It comprehensively summarizes the background related to FL and Blockchain, evaluates existing architectures for their integration, and the primary attacks and possible countermeasures to guarantee privacy in this setting. Finally, it reviews the main application scenarios where Blockchain-Enabled FL approaches have been proficiently applied. This survey can help academia and industry practitioners understand which theories and techniques exist to improve the performance of FL through Blockchain to preserve privacy and which are the main challenges and future directions in this novel and still under-explored context. We believe this work provides a novel contribution respect to the previous surveys and is a valuable tool to explore the current landscape, understand perspectives, and pave the way for advancements or improvements in this amalgamation of Blockchain and Federated Learning., Comment: 44 pages, 11 figures

Dynamic Wireless Power Transfer (DWPT) is a novel technology that allows charging an electric vehicle while driving thanks to a dedicated road infrastructure. DWPT's capabilities in automatically establishing charging sessions and billing without users' intervention make it prone to cybersecurity attacks. Hence, security is essential in preventing fraud, impersonation, and user tracking. To this aim, researchers proposed different solutions for authenticating users. However, recent advancements in quantum computing jeopardize classical public key cryptography, making currently existing solutions in DWPT authentication nonviable. To avoid the resource burden imposed by technology upgrades, it is essential to develop post-quantum-resistant solutions. In this paper, we propose DynamiQS, the first post-quantum secure authentication protocol for dynamic wireless charging. DynamiQS is privacy-preserving and secure against attacks on the DWPT. We leverage an Identity-Based Encryption with Lattices in the Ring Learning With Error framework. Furthermore, we show the possibility of using DynamiQS in a real environment, leveraging the results of cryptographic computation on real constrained devices and simulations. DynamiQS reaches a total time cost of around 281 ms, which is practicable in dynamic charging settings (car and charging infrastructure).

Decentralised learning has recently gained traction as an alternative to federated learning in which both data and coordination are distributed over its users. To preserve data confidentiality, decentralised learning relies on differential privacy, multi-party computation, or a combination thereof. However, running multiple privacy-preserving summations in sequence may allow adversaries to perform reconstruction attacks. Unfortunately, current reconstruction countermeasures either cannot trivially be adapted to the distributed setting, or add excessive amounts of noise. In this work, we first show that passive honest-but-curious adversaries can infer other users' private data after several privacy-preserving summations. For example, in subgraphs with 18 users, we show that only three passive honest-but-curious adversaries succeed at reconstructing private data 11.0% of the time, requiring an average of 8.8 summations per adversary. The success rate depends only on the adversaries' direct neighbourhood, independent of the size of the full network. We consider weak adversaries, who do not control the graph topology and can exploit neither the inner workings of the summation protocol nor the specifics of users' data. We develop a mathematical understanding of how reconstruction relates to topology and propose the first topology-based decentralised defence against reconstruction attacks. Specifically, we show that reconstruction requires a number of adversaries linear in the length of the network's shortest cycle. Consequently, reconstructing private data from privacy-preserving summations is impossible in acyclic networks. Our work is a stepping stone for a formal theory of topology-based reconstruction defences. Such a theory would generalise our countermeasure beyond summation, define confidentiality in terms of entropy, and describe the effects of differential privacy., Comment: 13 pages, 8 figures, submitted to PETS 2024, for associated experiment source code see doi:10.4121/21572601

Recently, we have witnessed a rise in the use of Large Language Models (LLMs), especially in applications like chatbot assistants. Safety mechanisms and specialized training procedures are implemented to prevent improper responses from these assistants. In this work, we bypass these measures for ChatGPT and Gemini (and, to some extent, Bing chat) by making them impersonate complex personas with personality characteristics that are not aligned with a truthful assistant. We start by creating elaborate biographies of these personas, which we then use in a new session with the same chatbots. Our conversations then follow a role-play style to elicit prohibited responses. Using personas, we show that prohibited responses are actually provided, making it possible to obtain unauthorized, illegal, or harmful information. This work shows that by using adversarial personas, one can overcome safety mechanisms set out by ChatGPT and Gemini. We also introduce several ways of activating such adversarial personas, which show that both chatbots are vulnerable to this kind of attack. With the same principle, we introduce two defenses that push the model to interpret trustworthy personalities and make it more robust against such attacks.

In the digital era, threat actors employ sophisticated techniques for which, often, digital traces in the form of textual data are available. Cyber Threat Intelligence~(CTI) is related to all the solutions inherent to data collection, processing, and analysis useful to understand a threat actor's targets and attack behavior. Currently, CTI is assuming an always more crucial role in identifying and mitigating threats and enabling proactive defense strategies. In this context, NLP, an artificial intelligence branch, has emerged as a powerful tool for enhancing threat intelligence capabilities. This survey paper provides a comprehensive overview of NLP-based techniques applied in the context of threat intelligence. It begins by describing the foundational definitions and principles of CTI as a major tool for safeguarding digital assets. It then undertakes a thorough examination of NLP-based techniques for CTI data crawling from Web sources, CTI data analysis, Relation Extraction from cybersecurity data, CTI sharing and collaboration, and security threats of CTI. Finally, the challenges and limitations of NLP in threat intelligence are exhaustively examined, including data quality issues and ethical considerations. This survey draws a complete framework and serves as a valuable resource for security professionals and researchers seeking to understand the state-of-the-art NLP-based threat intelligence techniques and their potential impact on cybersecurity.

Optical Character Recognition (OCR) is a widely used tool to extract text from scanned documents. Today, the state-of-the-art is achieved by exploiting deep neural networks. However, the cost of this performance is paid at the price of system vulnerability. For instance, in backdoor attacks, attackers compromise the training phase by inserting a backdoor in the victim's model that will be activated at testing time by specific patterns while leaving the overall model performance intact. This work proposes a backdoor attack for OCR resulting in the injection of non-readable characters from malicious input images. This simple but effective attack exposes the state-of-the-art OCR weakness, making the extracted text correct to human eyes but simultaneously unusable for the NLP application that uses OCR as a preprocessing step. Experimental results show that the attacked models successfully output non-readable characters for around 90% of the poisoned instances without harming their performance for the remaining instances.

Cyber Threat Intelligence (CTI) reporting is pivotal in contemporary risk management strategies. As the volume of CTI reports continues to surge, the demand for automated tools to streamline report generation becomes increasingly apparent. While Natural Language Processing techniques have shown potential in handling text data, they often struggle to address the complexity of diverse data sources and their intricate interrelationships. Moreover, established paradigms like STIX have emerged as de facto standards within the CTI community, emphasizing the formal categorization of entities and relations to facilitate consistent data sharing. In this paper, we introduce AGIR (Automatic Generation of Intelligence Reports), a transformative Natural Language Generation tool specifically designed to address the pressing challenges in the realm of CTI reporting. AGIR's primary objective is to empower security analysts by automating the labor-intensive task of generating comprehensive intelligence reports from formal representations of entity graphs. AGIR utilizes a two-stage pipeline by combining the advantages of template-based approaches and the capabilities of Large Language Models such as ChatGPT. We evaluate AGIR's report generation capabilities both quantitatively and qualitatively. The generated reports accurately convey information expressed through formal language, achieving a high recall value (0.99) without introducing hallucination. Furthermore, we compare the fluency and utility of the reports with state-of-the-art approaches, showing how AGIR achieves higher scores in terms of Syntactic Log-Odds Ratio (SLOR) and through questionnaires. By using our tool, we estimate that the report writing time is reduced by more than 40%, therefore streamlining the CTI production of any organization and contributing to the automation of several CTI tasks., Comment: 10 pages, 7 figures

While user authentication happens before initiating or resuming a login session, de-authentication detects the absence of a previously-authenticated user to revoke her currently active login session. The absence of proper de-authentication can lead to well-known lunchtime attacks, where a nearby adversary takes over a carelessly departed user's running login session. The existing solutions for automatic de-authentication have distinct practical limitations, e.g., extraordinary deployment requirements or high initial cost of external equipment. In this paper, we propose "DE-authentication using Ambient Light sensor" (DEAL), a novel, inexpensive, fast, and user-friendly de-authentication approach. DEAL utilizes the built-in ambient light sensor of a modern computer to determine if the user is leaving her work-desk. DEAL, by design, is resilient to natural shifts in lighting conditions and can be configured to handle abrupt changes in ambient illumination (e.g., due to toggling of room lights). We collected data samples from 4800 sessions with 120 volunteers in 4 typical workplace settings and conducted a series of experiments to evaluate the quality of our proposed approach thoroughly. Our results show that DEAL can de-authenticate a departing user within 4 seconds with a hit rate of 89.15% and a fall-out of 7.35%. Finally, bypassing DEAL to launch a lunchtime attack is practically infeasible as it requires the attacker to either take the user's position within a few seconds or manipulate the sensor readings sophisticatedly in real-time., Comment: This paper has been accepted for publication in IEEE ACCESS

To fight climate change, new "green" technology are emerging, most of them using electricity as a power source. Among the solutions, Electric Vehicles (EVs) represent a central asset in the future transport system. EVs require a complex infrastructure to enable the so-called Vehicle-to-Grid (V2G) paradigm to manage the charging process between the smart grid and the EV. In this paradigm, the Electric Vehicle Supply Equipment (EVSE), or charging station, is the end device that authenticates the vehicle and delivers the power to charge it. However, since an EVSE is publicly exposed and connected to the Internet, recent works show how an attacker with physical tampering and remote access can target an EVSE, exposing the security of the entire infrastructure and the final user. For this reason, it is important to develop novel strategies to secure such infrastructures. In this paper we present HoneyEVSE, the first honeypot conceived to simulate an EVSE. HoneyEVSE can simulate with high fidelity the EV charging process and, at the same time, enables a user to interact with it through a dashboard. Furthermore, based on other charging columns exposed on the Internet, we emulate the login and device information pages to increase user engagement. We exposed HoneyEVSE for 30 days to the Internet to assess its capability and measured the interaction received with its Shodan Honeyscore. Results show that HoneyEVSE can successfully evade the Shodan honeyscore metric while attracting a high number of interactions on the exposed services., Comment: 15 pages

Lithium-ion (Li-ion) batteries are the primary power source in various applications due to their high energy and power density. Their market was estimated to be up to 48 billion U.S. dollars in 2022. However, the widespread adoption of Li-ion batteries has resulted in counterfeit cell production, which can pose safety hazards to users. Counterfeit cells can cause explosions or fires, and their prevalence in the market makes it difficult for users to detect fake cells. Indeed, current battery authentication methods can be susceptible to advanced counterfeiting techniques and are often not adaptable to various cells and systems. In this paper, we improve the state of the art on battery authentication by proposing two novel methodologies, DCAuth and EISthentication, which leverage the internal characteristics of each cell through Machine Learning models. Our methods automatically authenticate lithium-ion battery models and architectures using data from their regular usage without the need for any external device. They are also resilient to the most common and critical counterfeit practices and can scale to several batteries and devices. To evaluate the effectiveness of our proposed methodologies, we analyze time-series data from a total of 20 datasets that we have processed to extract meaningful features for our analysis. Our methods achieve high accuracy in battery authentication for both architectures (up to 0.99) and models (up to 0.96). Moreover, our methods offer comparable identification performances. By using our proposed methodologies, manufacturers can ensure that devices only use legitimate batteries, guaranteeing the operational state of any system and safety measures for the users., Comment: 18 pages, 11 figures

Federated learning enables collaborative training of machine learning models by keeping the raw data of the involved workers private. Three of its main objectives are to improve the models' privacy, security, and scalability. Vertical Federated Learning (VFL) offers an efficient cross-silo setting where a few parties collaboratively train a model without sharing the same features. In such a scenario, classification labels are commonly considered sensitive information held exclusively by one (active) party, while other (passive) parties use only their local information. Recent works have uncovered important flaws of VFL, leading to possible label inference attacks under the assumption that the attacker has some, even limited, background knowledge on the relation between labels and data. In this work, we are the first (to the best of our knowledge) to investigate label inference attacks on VFL using a zero-background knowledge strategy. To formulate our proposal, we focus on Graph Neural Networks (GNNs) as a target model for the underlying VFL. In particular, we refer to node classification tasks, which are widely studied, and GNNs have shown promising results. Our proposed attack, BlindSage, provides impressive results in the experiments, achieving nearly 100% accuracy in most cases. Even when the attacker has no information about the used architecture or the number of classes, the accuracy remains above 90% in most instances. Finally, we observe that well-known defenses cannot mitigate our attack without affecting the model's performance on the main classification task.

Physical unclonable functions (PUFs) are hardware-oriented primitives that exploit manufacturing variations to generate a unique identity for a physical system. Recent advancements showed how DRAM can be exploited to implement PUFs. DRAM PUFs require no additional circuits for PUF operations and can be used in most of the applications with resource-constrained nodes such as Internet of Things (IoT) networks. However, the existing DRAM PUF solutions either require to interrupt other functions in the host system, or provide unreliable responses due to their sensitiveness to the environmental conditions. In this paper, we propose EPUF, a novel strategy to extract random and unique features from DRAM cells to generate reliable PUF responses. In particular, we use the bitmap images of the binary DRAM values and their entropy features. We show via real device experiments that EPUF is approximately $1.7$ times faster than other state of the art solutions, achieves $100\%$ reliability, generates features with $47.79\%$ uniqueness, and supports a large set of CRP that leads to new potentials for DRAM PUF-based authentication. We also propose a lightweight authentication protocol based on EPUF, which not only provides far better security guarantees but also outperforms the state-of-the-art in terms of communication overhead and computational cost.

Time-relevant and accurate threat information from public domains are essential for cyber security. In a constantly evolving threat landscape, such information assists security researchers in thwarting attack strategies. In this work, we collect and analyze threat-related information from Twitter to extract intelligence for proactive security. We first use a convolutional neural network to classify the tweets as containing or not valuable threat indicators. In particular, to gather threat intelligence from social media, the proposed approach collects pertinent Indicators of Compromise (IoCs) from tweets, such as IP addresses, URLs, File hashes, domain addresses, and CVE IDs. Then, we analyze the IoCs to confirm whether they are reliable and valuable for threat intelligence using performance indicators, such as correctness, timeliness, and overlap. We also evaluate how fast Twitter shares IoCs compared to existing threat intelligence services. Furthermore, through machine learning models, we classify Twitter accounts as either automated or human-operated and delve into the role of bot accounts in disseminating cyber threat information on social media. Our results demonstrate that Twitter is growing into a powerful platform for gathering precise and pertinent malware IoCs and a reliable source for mining threat intelligence.

Evasion attacks are a threat to machine learning models, where adversaries attempt to affect classifiers by injecting malicious samples. An alarming side-effect of evasion attacks is their ability to transfer among different models: this property is called transferability. Therefore, an attacker can produce adversarial samples on a custom model (surrogate) to conduct the attack on a victim's organization later. Although literature widely discusses how adversaries can transfer their attacks, their experimental settings are limited and far from reality. For instance, many experiments consider both attacker and defender sharing the same dataset, balance level (i.e., how the ground truth is distributed), and model architecture. In this work, we propose the DUMB attacker model. This framework allows analyzing if evasion attacks fail to transfer when the training conditions of surrogate and victim models differ. DUMB considers the following conditions: Dataset soUrces, Model architecture, and the Balance of the ground truth. We then propose a novel testbed to evaluate many state-of-the-art evasion attacks with DUMB; the testbed consists of three computer vision tasks with two distinct datasets each, four types of balance levels, and three model architectures. Our analysis, which generated 13K tests over 14 distinct attacks, led to numerous novel findings in the scope of transferable attacks with surrogate models. In particular, mismatches between attackers and victims in terms of dataset source, balance levels, and model architecture lead to non-negligible loss of attack performance., Comment: Accepted at RAID 2023

In February 2022, Russia launched a full-scale invasion of Ukraine. This event had global repercussions, especially on the political decisions of European countries. As expected, the role of Italy in the conflict became a major campaign issue for the Italian General Election held on 25 September 2022. Politicians frequently use Twitter to communicate during political campaigns, but bots often interfere and attempt to manipulate elections. Hence, understanding whether bots influenced public opinion regarding the conflict and, therefore, the elections is essential. In this work, we investigate how Italian politics responded to the Russo-Ukrainian conflict on Twitter and whether bots manipulated public opinion before the 2022 general election. We first analyze 39,611 tweets of six major political Italian parties to understand how they discussed the war during the period February-December 2022. Then, we focus on the 360,823 comments under the last month's posts before the elections, discovering around 12% of the commenters are bots. By examining their activities, it becomes clear they both distorted how war topics were treated and influenced real users during the last month before the elections.

Many research papers have recently focused on behavioral-based driver authentication systems in vehicles. Pushed by Artificial Intelligence (AI) advancements, these works propose powerful models to identify drivers through their unique biometric behavior. However, these models have never been scrutinized from a security point of view, rather focusing on the performance of the AI algorithms. Several limitations and oversights make implementing the state-of-the-art impractical, such as their secure connection to the vehicle's network and the management of security alerts. Furthermore, due to the extensive use of AI, these systems may be vulnerable to adversarial attacks. However, there is currently no discussion on the feasibility and impact of such attacks in this scenario. Driven by the significant gap between research and practical application, this paper seeks to connect these two domains. We propose the first security-aware system model for behavioral-based driver authentication. We develop two lightweight driver authentication systems based on Random Forest and Recurrent Neural Network architectures designed for our constrained environments. We formalize a realistic system and threat model reflecting a real-world vehicle's network for their implementation. When evaluated on real driving data, our models outclass the state-of-the-art with an accuracy of up to 0.999 in identification and authentication. Moreover, we are the first to propose attacks against these systems by developing two novel evasion attacks, SMARTCAN and GANCAN. We show how attackers can still exploit these systems with a perfect attack success rate (up to 1.000). Finally, we discuss requirements for deploying driver authentication systems securely. Through our contributions, we aid practitioners in safely adopting these systems, help reduce car thefts, and enhance driver security.

Recently, researchers have successfully employed Graph Neural Networks (GNNs) to build enhanced recommender systems due to their capability to learn patterns from the interaction between involved entities. In addition, previous studies have investigated federated learning as the main solution to enable a native privacy-preserving mechanism for the construction of global GNN models without collecting sensitive data into a single computation unit. Still, privacy issues may arise as the analysis of local model updates produced by the federated clients can return information related to sensitive local data. For this reason, experts proposed solutions that combine federated learning with Differential Privacy strategies and community-driven approaches, which involve combining data from neighbor clients to make the individual local updates less dependent on local sensitive data. In this paper, we identify a crucial security flaw in such a configuration, and we design an attack capable of deceiving state-of-the-art defenses for federated learning. The proposed attack includes two operating modes, the first one focusing on convergence inhibition (Adversarial Mode), and the second one aiming at building a deceptive rating injection on the global federated model (Backdoor Mode). The experimental results show the effectiveness of our attack in both its modes, returning on average 60% performance detriment in all the tests on Adversarial Mode and fully effective backdoors in 93% of cases for the tests performed on Backdoor Mode.

Search engines are vulnerable to attacks against indexing and searching via text encoding manipulation. By imperceptibly perturbing text using uncommon encoded representations, adversaries can control results across search engines for specific search queries. We demonstrate that this attack is successful against two major commercial search engines - Google and Bing - and one open source search engine - Elasticsearch. We further demonstrate that this attack is successful against LLM chat search including Bing's GPT-4 chatbot and Google's Bard chatbot. We also present a variant of the attack targeting text summarization and plagiarism detection models, two ML tasks closely tied to search. We provide a set of defenses against these techniques and warn that adversaries can leverage these attacks to launch disinformation campaigns against unsuspecting users, motivating the need for search engine maintainers to patch deployed systems., Comment: To appear in the 26th Symposium on Research in Attacks, Intrusions and Defenses (RAID). Revisions: Adds table summarizing attacks

The capability of doing effective forensic analysis on printed and scanned (PS) images is essential in many applications. PS documents may be used to conceal the artifacts of images which is due to the synthetic nature of images since these artifacts are typically present in manipulated images and the main artifacts in the synthetic images can be removed after the PS. Due to the appeal of Generative Adversarial Networks (GANs), synthetic face images generated with GANs models are difficult to differentiate from genuine human faces and may be used to create counterfeit identities. Additionally, since GANs models do not account for physiological constraints for generating human faces and their impact on human IRISes, distinguishing genuine from synthetic IRISes in the PS scenario becomes extremely difficult. As a result of the lack of large-scale reference IRIS datasets in the PS scenario, we aim at developing a novel dataset to become a standard for Multimedia Forensics (MFs) investigation which is available at [45]. In this paper, we provide a novel dataset made up of a large number of synthetic and natural printed IRISes taken from VIPPrint Printed and Scanned face images. We extracted irises from face images and it is possible that the model due to eyelid occlusion captured the incomplete irises. To fill the missing pixels of extracted iris, we applied techniques to discover the complex link between the iris images. To highlight the problems involved with the evaluation of the dataset's IRIS images, we conducted a large number of analyses employing Siamese Neural Networks to assess the similarities between genuine and synthetic human IRISes, such as ResNet50, Xception, VGG16, and MobileNet-v2. For instance, using the Xception network, we achieved 56.76\% similarity of IRISes for synthetic images and 92.77% similarity of IRISes for real images.

Social Honeypots are tools deployed in Online Social Networks (OSN) to attract malevolent activities performed by spammers and bots. To this end, their content is designed to be of maximum interest to malicious users. However, by choosing an appropriate content topic, this attractive mechanism could be extended to any OSN users, rather than only luring malicious actors. As a result, honeypots can be used to attract individuals interested in a wide range of topics, from sports and hobbies to more sensitive subjects like political views and conspiracies. With all these individuals gathered in one place, honeypot owners can conduct many analyses, from social to marketing studies. In this work, we introduce a novel concept of social honeypot for attracting OSN users interested in a generic target topic. We propose a framework based on fully-automated content generation strategies and engagement plans to mimic legit Instagram pages. To validate our framework, we created 21 self-managed social honeypots (i.e., pages) on Instagram, covering three topics, four content generation strategies, and three engaging plans. In nine weeks, our honeypots gathered a total of 753 followers, 5387 comments, and 15739 likes. These results demonstrate the validity of our approach, and through statistical analysis, we examine the characteristics of effective social honeypots., Comment: Accepted at ACNS2023