Privacy Risks Research Articles

The Use of AI in Medicine: Health Data, Privacy Risks and More

In the era of advancements in artificial intelligence (AI) and machine learning, the healthcare industry has become one of the major areas where such technologies are being actively adopted and utilized. The global health care sector generated more than 2.3 zettabytes of data worldwide in 2020. Analysts estimate that the global market for artificial intelligence (AI) in medicine will grow to $13 billion by 2025, with a significant increase in newly established companies. Artificial intelligence in medicine is used to predict, detect and diagnose various diseases and pathologies. The sources of data can be various results of medical research (EEG, X-ray images, laboratory tests, e.g. tissues, etc.). At the same time, there are understandable concerns that AI will undermine the patient-provider relationship, contribute to the deskilling of providers, undermine transparency, misdiagnose or inappropriately treat because of errors within AI decision-making that are hard to detect, exacerbate existing racial or societal biases, or introduce algorithmic bias that will be hard to detect. Traditional research methods, general and special ones, with an emphasis on the comparative legal method, were chosen. For the AI to work it needs to be trained, and it’s learning from all sorts of information given to it. The main part of the information on which AI is trained is health data, which is sensitive personal data. The fact that personal data is qualified as sensitive personal data indicates the significance of the information contained, the high risks in case it’s leaking, and hence the need for stricter control and regulation. The article offers a detailed exploration of the legal implications of AI in medicine, highlighting existing challenges, the current state of regulation, and proposes future perspectives and recommendations for legislation adapted to the era of medical AI. Given the above, the study is divided into three parts: international framework, that will focus primarily on applicable WHO documents; risks and possible ways to minimize them, where the authors have tried to consider various issues related to the use of AI in medicine and find options to address them; and relevant case-study.

Hybrid quantum architecture for smart city security

Currently and in the near future, Smart Cities are vital to enhance urban living, address resource challenges, optimize infrastructure, and harness technology for sustainability, efficiency, and improved quality of life in rapidly urbanizing environments. Owing to the high usage of networks, sensors, and connected devices, Smart Cities generate a massive amount of data. Therefore, Smart City security concerns encompass data privacy, Internet-of-Things (IoT) vulnerabilities, cyber threats, and urban infrastructure risks, requiring robust solutions to safeguard digital assets, citizens, and critical services. Some solutions include robust cybersecurity measures, data encryption, Artificial Intelligence (AI)-driven threat detection, public–private partnerships, standardized security protocols, and community engagement to foster a resilient and secure smart city ecosystem. For example, Security Information and Event Management (SIEM) helps in real-time monitoring, threat detection, and incident response by aggregating and analyzing security data. To this end, no integrated systems are operating in this context. In this paper, we propose a Hybrid Quantum-Classical Architecture for bolstering Smart City security that exploits Quantum Machine Learning (QML) and SIEM to provide security based on Quantum Artificial Intelligence and patterns/rules. The validity of the hybrid quantum-classical architecture was proven by conducting experiments and a comparison of the QML algorithms with state-of-the-art AI algorithms. We also provide a proof of concept dashboard for the proposed architecture.

Anonymization: The imperfect science of using data while preserving privacy.

Information about us, our actions, and our preferences is created at scale through surveys or scientific studies or as a result of our interaction with digital devices such as smartphones and fitness trackers. The ability to safely share and analyze such data is key for scientific and societal progress. Anonymization is considered by scientists and policy-makers as one of the main ways to share data while minimizing privacy risks. In this review, we offer a pragmatic perspective on the modern literature on privacy attacks and anonymization techniques. We discuss traditional de-identification techniques and their strong limitations in the age of big data. We then turn our attention to modern approaches to share anonymous aggregate data, such as data query systems, synthetic data, and differential privacy. We find that, although no perfect solution exists, applying modern techniques while auditing their guarantees against attacks is the best approach to safely use and share data today.

Policy Guidance for Direct-to-Consumer Genetic Testing Services: Framework Development Study.

The online offer of commercial genetic tests, also called direct-to-consumer genetic tests (DTC-GTs), enables citizens to gain insight into their health and disease risk based on their genetic profiles. DTC-GT offers often consist of a combination of services or aspects, including advertisements, information, DNA analysis, and medical or lifestyle advice. The risks and benefits of DTC-GT services have been debated and studied extensively, but instruments that assess DTC-GT services and aid policy are lacking. This leads to uncertainty among policy makers, law enforcers, and regulators on how to ensure and balance both public safety and autonomy and about the responsibilities these 3 parties have toward the public. This study aimed to develop a framework that outlines aspects of DTC-GTs that lead to policy issues and to help provide policy guidance regarding DTC-GT services. We performed 3 steps: (1) an integrative literature review to identify risks and benefits of DTC-GT services for consumers and society in Embase and Medline (January 2014-June 2022), (2) structuring benefits and risks in different steps of the consumer journey, and (3) development of a checklist for policy guidance. Potential risks and benefits of DTC-GT services were mapped from 134 papers and structured into 6 phases. In summary, these phases were called the consumer journey: (1) exposure, (2) pretest information, (3) DNA analysis, (4) data management, (5) posttest information, and (6) individual and societal impact. The checklist for evaluation of DTC-GT services consisted of 8 themes, covering 38 items that may raise policy issues in DTC-GT services. The themes included the following aspects: general service content, validity and quality assurance, potential data and privacy risks, scientific evidence and robustness, and quality of the provided information. Both the consumer journey and the checklist break the DTC-GT offer down into key aspects that may impact and compromise individual and public health, safety, and autonomy. This framework helps policy makers, regulators, and law enforcers develop methods to interpret, assess, and act in the DTC-GT service market.

Unveiling barriers to O2O technology platform adoption among small retailers in India: insights into the role of digital ecosystem

Purpose This study aims to delve into the intricate world of small retailers in India, seeking to understand the impediments/barriers they encounter when trying to embrace online-to-offline (O2O) platforms. It also investigates the potential impact of the digital ecosystem in moderating these barriers. Design/methodology/approach Data from 426 Indian retailers was collected, and structural equation modelling was used to validate the conceptual framework. Findings The findings highlight the importance of addressing distrust and technological anxiety as key barriers to O2O platform adoption. Psychological risk, low-tech orientation, privacy risk, financial risk and social risk were also identified as barriers. Interestingly, performance risk and infrastructure were found to be insignificant in this study. The study indicates that the digital ecosystem does moderate the relationship between psychological risk, performance risk, distrust and technological anxiety with attitude. Research limitations/implications This research holds significant implications for technology adoption, retail management and aggregator platform development in developing nations, notably India. This research draws upon a conceptual framework to deepen the understanding of the O2O technology platform by providing an all-inclusive overview. Originality/value This study breaks new ground by investigating the distinctive obstacles to O2O adoption faced by small retailers in India. By validating the digital ecosystem’s moderating effect, this research yields insights that are context-specific and particularly relevant to the Indian retail landscape. Valuable guidance is offered for researchers, practitioners and policymakers navigating O2O strategy implementation in emerging markets.

BFLIDS: Blockchain-Driven Federated Learning for Intrusion Detection in IoMT Networks.

The Internet of Medical Things (IoMT) has significantly advanced healthcare, but it has also brought about critical security challenges. Traditional security solutions struggle to keep pace with the dynamic and interconnected nature of IoMT systems. Machine learning (ML)-based Intrusion Detection Systems (IDS) have been increasingly adopted to counter cyberattacks, but centralized ML approaches pose privacy risks due to the single points of failure (SPoFs). Federated Learning (FL) emerges as a promising solution, enabling model updates directly on end devices without sharing private data with a central server. This study introduces the BFLIDS, a Blockchain-empowered Federated Learning-based IDS designed to enhance security and intrusion detection in IoMT networks. Our approach leverages blockchain to secure transaction records, FL to maintain data privacy by training models locally, IPFS for decentralized storage, and MongoDB for efficient data management. Ethereum smart contracts (SCs) oversee and secure all interactions and transactions within the system. We modified the FedAvg algorithm with the Kullback-Leibler divergence estimation and adaptive weight calculation to boost model accuracy and robustness against adversarial attacks. For classification, we implemented an Adaptive Max Pooling-based Convolutional Neural Network (CNN) and a modified Bidirectional Long Short-Term Memory (BiLSTM) with attention and residual connections on Edge-IIoTSet and TON-IoT datasets. We achieved accuracies of 97.43% (for CNNs and Edge-IIoTSet), 96.02% (for BiLSTM and Edge-IIoTSet), 98.21% (for CNNs and TON-IoT), and 97.42% (for BiLSTM and TON-IoT) in FL scenarios, which are competitive with centralized methods. The proposed BFLIDS effectively detects intrusions, enhancing the security and privacy of IoMT networks.

A Conditional Privacy-Preserving Identity-Authentication Scheme for Federated Learning in the Internet of Vehicles.

With the rapid development of artificial intelligence and Internet of Things (IoT) technologies, automotive companies are integrating federated learning into connected vehicles to provide users with smarter services. Federated learning enables vehicles to collaboratively train a global model without sharing sensitive local data, thereby mitigating privacy risks. However, the dynamic and open nature of the Internet of Vehicles (IoV) makes it vulnerable to potential attacks, where attackers may intercept or tamper with transmitted local model parameters, compromising their integrity and exposing user privacy. Although existing solutions like differential privacy and encryption can address these issues, they may reduce data usability or increase computational complexity. To tackle these challenges, we propose a conditional privacy-preserving identity-authentication scheme, CPPA-SM2, to provide privacy protection for federated learning. Unlike existing methods, CPPA-SM2 allows vehicles to participate in training anonymously, thereby achieving efficient privacy protection. Performance evaluations and experimental results demonstrate that, compared to state-of-the-art schemes, CPPA-SM2 significantly reduces the overhead of signing, verification and communication while achieving more security features.

ST-TrajGAN: A synthetic trajectory generation algorithm for privacy preservation

The rapid growth of large-scale trajectory data poses privacy risks for location-based services (LBS), primarily through centralized storage and processing of data, as well as insecure data transmission channels (such as the Internet and wireless networks), which can lead to unauthorized access or manipulation of users' location information by attackers. To enhance trajectory privacy protection while improving the trajectory utility, this paper proposes an efficient and secure deep learning model Semantic and Transformer-based Trajectory Generative Adversarial Networks (ST-TrajGAN) for trajectory data generation and publication. First, this article introduces a semantic trajectory encoding model for preprocessing trajectory points. Through this model, trajectory points can be transformed into vector representations with semantic information. Next, by learning the spatio-temporal and semantic features of real trajectory data, a deep learning model is used to generate synthetic trajectories with more uncertainty and practicality. Furthermore, a novel TrajLoss loss metric function was crafted to gauge the trajectory similarity loss within the trained deep learning model. Ultimately, the efficacy of the generated synthetic trajectories and the model's utility are assessed through Trajectory-User Linking (TUL) and Trajectory Sharing Percentage (TSP) values on three authentic Location-Based Services (LBS) datasets. Numerous experiments have shown that our method outperforms other methods in terms of privacy protection effectiveness and utility.

How consumers' privacy perceptions influenced mobile payment acceptance during the COVID-19 pandemic

PurposeOur research investigated how the perceived effectiveness of privacy, perceived privacy risk, and perceived security influenced consumers' behavioral intention to use mobile payment applications during the COVID-19 pandemic.Design/methodology/approachWe applied a quantitative method using a cross-sectional online survey conducted over three years. We collected a sample of 1,471 survey responses focused on ages 18–39. Using descriptive statistics, confirmatory factor analysis, and structural equation modeling, we tested our hypotheses with SPSS 27 and AMOS 27.FindingsResults of the study indicate that the perceived effectiveness of privacy positively influences perceived privacy risk, perceived security, and behavioral intention. Moreover, perceived privacy risk has a positive effect on perceived security. We found no significant relationship between perceived privacy risk and behavioral intention, although perceived security has a positive effect on behavioral intention. Further mediation analyses showed that perceived privacy risk and perceived security mediate the relationship between the perceived effectiveness of privacy and behavioral intention.Originality/valueThis research sheds new light on the role of perceived privacy effectiveness in mobile payment adoption in Hungary, particularly during the COVID-19 pandemic. Our research also explains why and how perceived privacy effectiveness influences consumers' perceived privacy risk, perceived security, and behavioral intention.

Challenges of Using Synthetic Data Generation Methods for Tabular Microdata

The generation of synthetic data holds significant promise for augmenting limited datasets while avoiding privacy issues, facilitating research, and enhancing machine learning models’ robustness. Generative Adversarial Networks (GANs) stand out as promising tools, employing two neural networks—generator and discriminator—to produce synthetic data that mirrors real data distributions. This study evaluates GAN variants (CTGAN, CopulaGAN), a variational autoencoder, and copulas on diverse real datasets of different complexity encompassing numerical and categorical attributes. The results highlight CTGAN’s sensitivity to training parameters and TVAE’s robustness across datasets. Scalability challenges persist, with GANs demanding substantial computational resources. TVAE stands out for its high utility across all datasets, even for high-dimensional data, though it incurs higher privacy risks, which is indicative of the curse of dimensionality. While no single model universally excels, understanding the trade-offs and leveraging model strengths can significantly enhance synthetic data generation (SDG). Future research should focus on adaptive learning mechanisms, scalability enhancements, and standardized evaluation metrics to advance SDG methods effectively. Addressing these challenges will foster broader adoption and application of synthetic data.

Does CLIP Know My Face?

With the rise of deep learning in various applications, privacy concerns around the protection of training data have become a critical area of research. Whereas prior studies have focused on privacy risks in single-modal models, we introduce a novel method to assess privacy for multi-modal models, specifically vision-language models like CLIP. The proposed Identity Inference Attack (IDIA) reveals whether an individual was included in the training data by querying the model with images of the same person. Letting the model choose from a wide variety of possible text labels, the model reveals whether it recognizes the person and, therefore, was used for training. Our large-scale experiments on CLIP demonstrate that individuals used for training can be identified with very high accuracy. We confirm that the model has learned to associate names with depicted individuals, implying the existence of sensitive information that can be extracted by adversaries. Our results highlight the need for stronger privacy protection in large-scale models and suggest that IDIAs can be used to prove the unauthorized use of data for training and to enforce privacy laws. This article appears in the AI & Society track.

A review of privacy-preserving research on federated graph neural networks

Graph neural networks are widely employed in diverse domains; however, they confront the peril of privacy infringement. To address this concern, federated learning emerges as a privacy-preserving approach that avoids sharing data for model training, effectively resolving the issue of privacy leakage in graph neural networks. The rapid advancement of federated neural networks has spurred the demand for more potent tools to enhance model performance owing to the concealed correlation information amongst federated learning participants. However, the structural attributes of federated graph neural networks render them vulnerable to inference attacks, reconstruction attacks, inversion attacks, and the like, potentially endangering privacy. This study delves into the intricacies of privacy-preserving within federated graph neural networks. Firstly, it introduces the architecture and variants of federated graph neural networks, analyzes the privacy risks encountered by these networks from four perspectives, and elucidates three primary attack methods. In accordance with the privacy-preserving mechanism of federated graph neural networks, it summarizes the privacy-preserving techniques and synthesizes the existing strategies from four perspectives: encryption methods, perturbation methods, anonymization, and hybrid methods. Furthermore, it summarily presents the prevailing framework for preserving privacy in neural networks. Ultimately, this paper examines the challenges and outlines future research directions pertaining to federated graph neural network technology.

Digital Entrepreneurship, The Risk and Business Viability Assessment in Nigeria

The study assessed digital entrepreneurship, the risk and the business viability in Nigeria. The study's specific goal is to ascertain the degree of viability and risk assessment of digital entrepreneurship in Nigeria’s economic space. Digital entrepreneurship is a 21st century business model because it is an enterprise that relies only on a digital platform that is pursued and managed by a digital entrepreneur. Digital entrepreneurs rely on internet resources to create and expand their digital services enterprises. The study's sample included 400 respondents who were chosen from Delta State's small and medium-sized business owners and proprietors. The adopted descriptive survey method and simple percentages were used to evaluate the data. The overall conclusion of the study is that digital entrepreneurship is a highly effective and viable business model in Nigeria and it is fantastic in bringing goods and services to customers' doorsteps, provides business owners with a platform for brand development and content creation as well as business growth. The study identified five major risks associated with digital entrepreneurship as cyber security risk, information security and privacy risk, fraud risk, copyright violation risk, infringement of trademark risk. The study confirmed that digital entrepreneurship creates business values at profit with lower cost of production when compared to the cost of operating physically established businesses or other means of doing business offline. The study also concludes empirically that digital entrepreneurship has a low possibility of loss but high profit possibility with the ability to manage the four identified risks.

Reducing privacy risks of China's healthcare big data through the policy framework.

While healthcare big data brings great opportunities and convenience to the healthcare industry, it also inevitably raises the issue of privacy leakage. Nowadays, the whole world is facing the security threat of healthcare big data, for which a sound policy framework can help reduce privacy risks of healthcare big data. In recent years, the Chinese government and industry self-regulatory organizations have issued a series of policy documents to reduce privacy risks of healthcare big data. However, China's policy framework suffers from the drawbacks of the mismatched operational model, the inappropriate operational method, and the poorly actionable operational content. Based on the experiences of the European Union, Australia, the United States, and other extra-territorial regions, strategies are proposed for China to amend the operational model of the policy framework, improve the operational method of the policy framework, and enhance the operability of the operational content of the policy framework. This study enriches the research on China's policy framework to reduce privacy risks of healthcare big data and provides some inspiration for other countries.

Security risks of chatbots in customer service: a comprehensive literature review

Chatbots are a powerful tool that are completely transforming customer service. They not only provide round-the-clock support, they also personalize every interaction. However, they present significant security and privacy risks. This comprehensive literature review aims to determine the vulnerabilities and threats associated with chatbots in customer service, and identifies effective mitigation strategies, including robust authentication mechanisms, end-to-end encryption, regular security audits, the implementation of security development lifecycles, MAC address authentication, and the use of blockchain technology.

Coping Responses to the Stress of Using News Platforms’ Recommendation Algorithms

Drawing on the stressor–strain–outcome (SSO) framework, this study examined how information-related stressors (perceived information overload, information quality deterioration, and information narrowing) and privacy-related stressor (perceived privacy risk) in using news platforms’ recommendation algorithms trigger users’ psychological reactance, which leads to their active coping, expressive coping, or avoidance coping with their stress in using such recommendation algorithms. Using a sample of 352 valid research responses, this study found that users’ psychological reactance is positively influenced by perceived information quality deterioration, followed by perceived information overload and perceived privacy risk. Meanwhile, perceived information narrowing has no significant impact on psychological reactance. In turn, psychological reactance exerts a stronger impact on expressive coping and avoidance coping responses than on active coping response. This study enriches the literature on the stress in algorithm usage at the individual level and provides practical insights for the providers of news platforms which have applied recommendation algorithms.

Industrial Metaverse: A Comprehensive Review, Environmental Impact, and Challenges

The Industrial Metaverse paradigm can be broadly described as a virtual environment that integrates various technologies such as augmented reality and mixed reality to enhance business operations and processes. It aims to streamline workflows, reduce error rates, improve efficiency, and provide a more engaging experience for employees. The promise of the Industrial Metaverse to drive sustainability and resource efficiency is compelling. Using advanced technologies such as the Industrial Metaverse is vital in an endeavor to have a competitive edge in a rapidly evolving business environment. However, the environmental impact of the technologies underpinning the Industrial Metaverse, like data centers and network infrastructure, should not be overlooked. The ecological footprint of these technologies must be considered in the sustainability equation. Researchers have warned that, by 2025, without sustainable artificial intelligence (AI) practices, AI will consume more energy than the human workforce, significantly offsetting zero carbon gains. As the Metaverse persists in evolving and gaining momentum, it will be necessary for companies to prioritize sustainability and explore new ways to balance technological advancements with environmental stewardship. However, recent studies have conjectured that the Metaverse holds the potential to reduce carbon emissions, as digital replacements for physical goods become more prevalent and physical activities like mobility and construction are reduced. Moreover, the specific extent to which this substitution can alleviate environmental concerns remains an open issue, presenting a knowledge gap in understanding the real-world impact of digital replacements. Thus, the objective of this paper is to provide a comprehensive review of the Industrial Metaverse, as well as explore the environmental impact of the Industrial Metaverse. The integrative literature review design and methodological approach involved multiple sources from the Web of Science and databases such as the ACM library, IEEE Library, and Google Scholar, which were analyzed to provide a comprehensive understanding of the developments in the Industrial Metaverse. Firstly, by considering the Industrial Metaverse’s architecture, we elucidate the Industrial Metaverse concept and the associated enabling technologies. Secondly, we performed an exploration through a discussion of the prevalent use cases and the deployment of the emerging Industrial Metaverse. Thirdly, we explored the impact of the Industrial Metaverse on the environment. Lastly, we address novel security and privacy risks, as well as upcoming research challenges, keeping in mind that the Industrial Metaverse is based on a strong data fabric. The results point to the Industrial Metaverse as having both positive and negative environmental effects in terms of energy consumption, e-waste, and pollution. Research, however, indicates that most Industrial Metaverse applications have a positive environmental impact and subsequently trend toward sustainability. Finally, for sustainability in the Industrial Metaverse, enterprises may consider utilizing renewable energy sources and cloud services. Furthermore, we examined the effects of products on the environment, as well as in the creation of a circular economy.

PrivySeC: A secure and privacy-compliant distributed framework for personal data sharing in IoT ecosystems

The contemporary era experiences an unprecedented dependence on data generated by individuals via an array of interconnected devices constituting the Internet of Things (IoT). The information amassed through IoT devices serves many objectives, including prescriptive analytics and predictive maintenance, preemptive healthcare measures, disaster mitigation, operational efficiency, and increased yield. In contrast, most applications or systems that rely on user-generated data to fulfill their business objectives face challenges in adhering to privacy protocols. Consequently, users are exposed to many privacy risks. Such infringements upon privacy provisions give rise to apprehensions regarding the authenticity of the processed data. Hence, this paper presents weaknesses and challenges in current practices and proposes “PrivySeC,” a distributed ledger technology (DLT) based framework for privacy-preserving and secure sharing of personally and non-personally identifiable information. The security analysis indicates that the proposed solution ensures data privacy by design and complies with most of the requirements mandated by various privacy regulations. Similarly, PrivySeC promises low transaction latency and provides high throughput. Although we have created a privacy-preserving solution for sharing smart farm data, it can be customized to meet the specific privacy requirements of individual applications.

Exploring Design Opportunities for Family-Based Privacy Education in Informal Learning Spaces

Children face increasing privacy risks and the need to navigate complex choices, while privacy education is not sufficient due to limited education scope and family involvement. We advocate for informal learning spaces (ILS) as a pioneering channel for family-based privacy education, given their established role in holistic technology and digital literacy education, which specifically targets family groups. In this paper, we conducted an interview study with eight families to understand revealing current approaches to privacy education and engagement with ILS for family-based learning. Our findings highlight ILS’s transformative potential in family privacy education, considering existing practices and challenges. We discuss the design opportunities for family-based privacy education in ILS, covering goals, content, engagement, and experience design. These insights contribute to future research on family-based privacy education in ILS.

GCL-Leak: Link Membership Inference Attacks against Graph Contrastive Learning

Graph contrastive learning (GCL) has emerged as a successful method for self-supervised graph learning. It involves generating augmented views of a graph by augmenting its edges and aims to learn node embeddings that are invariant to graph augmentation. Despite its effectiveness, the potential privacy risks associated with GCL models have not been thoroughly explored. In this paper, we delve into the privacy vulnerability of GCL models through the lens of link membership inference attacks (LMIA). Specifically, we focus on the federated setting where the adversary has white-box access to the node embeddings of all the augmented views generated by the target GCL model. Designing such white-box LMIAs against GCL models presents a significant and unique challenge due to potential variations in link memberships among node pairs in the target graph and its augmented views. This variability renders members indistinguishable from non-members when relying solely on the similarity of their node embeddings in the augmented views. To address this challenge, our in-depth analysis reveals that the key distinguishing factor lies in the similarity of node embeddings within augmented views where the node pairs share identical link memberships as those in the training graph. However, this poses a second challenge, as information about whether a node pair has identical link membership in both the training graph and augmented views is only available during the attack training phase. This demands the attack classifier to handle the additional “identical-membership" information which is available only for training and not for testing. To overcome this challenge, we propose GCL-LEAK, the first link membership inference attack against GCL models. The key component of GCL-LEAK is a new attack classifier model designed under the “Learning Using Privileged Information (LUPI)” paradigm, where the privileged information of “same-membership” is encoded as part of the attack classifier's structure. Our extensive set of experiments on four representative GCL models showcases the effectiveness of GCL-LEAK. Additionally, we develop two defense mechanisms that introduce perturbation to the node embeddings. Our empirical evaluation demonstrates that both defense mechanisms significantly reduce attack accuracy while preserving the accuracy of GCL models.