ABSTRACT Around the world, there are growing concerns about the harms youth experience on social networking platforms. However, the question of whether to restrict young people’s access to social media through legislative or policy mechanisms is widely contested. Understanding how the media portrays this debate is critical because the framing of policies, problems, supporters, and opponents can be powerful in shaping public opinion of youth and youth issues. This study aimed to examine how debate around Australia’s Social Media Minimum Age law was framed in Australian and Aotearoa New Zealand media coverage. Applying reflexive thematic analysis to 58 media texts located through a systematic search showed coverage was framed around four key themes: Protecting wellbeing; Protection from harmful content and actions; Protection for parents; and Protection of privacy. Together, these themes position youth as passive, vulnerable, and in need of protections, reinforcing dominant social constructions of youth and raising questions about how young people’s agency is framed and empowered in public discourse. Discussion argues efforts to improve youth wellbeing require moving beyond ‘moral panics’ and ‘quick fixes’ to co-creating policy solutions with youth, in ways that account for the broader social, economic, and political dimensions that shape young people’s lived experiences.

Artificial intelligence (AI) is rapidly shifting from experimental pilots to mainstream clinical infrastructure, redefining how evidence, accountability, and ethics intersect in healthcare. This narrative review integrates insights from peer-reviewed studies and policy frameworks to examine seven cross-cutting aspects: bias and fairness, explainability, safety and quality, privacy and data protection, accountability and liability, human oversight, and procurement and deployment. Findings reveal persistent inequities driven by dataset bias and opaque design; the need for explainability tools that are validated, task-specific, and usable by clinicians; and the centrality of post-market surveillance for sustaining patient safety. Privacy-preserving methods such as federated learning and differential privacy show promise but demand rigorous validation and regulatory coherence. Emerging liability models advocate shared enterprise responsibility, while governance-by-design—embedding transparency, auditability, and equity across the AI lifecycle—appears most effective in balancing innovation with public trust. Ethical, legal, and technical safeguards must evolve together to ensure that AI augments, rather than replaces, clinical judgment and institutional accountability.

As social media platforms increasingly monetize user data through targeted advertising, critical questions arise about privacy rights, digital commodification, and platform governance. This study examines how Pinterest users conceptualize and value privacy, ad-free experiences, and alternative platform ownership models, including subscription-based and cooperative structures. Through mixed methods analysis of 1,000 Pinterest users’ responses, we investigate willingness to pay (WTP) for enhanced privacy protections and data sovereignty. Quantitative analysis reveals that revenue-sharing beliefs (β = 1.17, p < .001), privacy concerns (β = 0.29, p < .001), and income (β = 0.27, p < .001) significantly predict WTP, while age shows a negative association (β = -0.45, p < .001). Qualitative findings illuminate the mechanisms underlying these patterns, revealing tensions between users’ stated privacy concerns and their behavioral practices, extending scholarship on the privacy paradox. Although respondents demonstrate awareness of their uncompensated digital labor, structural barriers temper enthusiasm for alternative models. These results advance platform studies and digital sociology by illuminating the complex interplay between surveillance capitalism, user agency, and economic constraints. The study concludes by discussing practical implications for platform design, policy development, and future research on digital rights and platform sustainability.

Simulation-based education is widely used to enhance clinical competence and communication skills in health professions. However, limited evidence exists on combined educational models that integrate simulation with structured counseling to strengthen empathy and privacy protection among midwifery students. This randomized controlled trial examined the effects of a two-component educational intervention-wearable birth simulation followed by seven weeks of structured counseling-on students' empathic tendency and privacy protection competencies. A total of 105 final-year midwifery students were randomly assigned to three groups: normal birth simulation (n = 35), breech birth simulation (n = 35), and control (n = 35). Data were collected at four time points using the Empathic Tendency Scale and the Privacy Protection in Obstetrics and Gynecology Scale. Mixed-model analyses were applied. Students in the breech birth simulation group demonstrated significantly higher empathy scores compared with the control group (p = 0.025). Across all groups, empathy and privacy awareness increased significantly over time (p < 0.001). However, the intervention did not produce significant improvements in the frequency of applying privacy-protective behaviors (p > 0.05). The combined intervention-simulation followed by structured counseling-enhanced students' empathy and awareness regarding privacy protection, although behavioral change did not occur to the same extent. Integrating multi-component educational strategies into midwifery curricula may strengthen both emotional and cognitive dimensions of respectful maternity care. ClinicalTrials.gov, NCT05864859. Registered on 05 May 2023.

Abstract Introduction Literature reviews indicate older adults are less engaged in using digital technologies due to reasons such as fear of falling and perceived lack of time. However, there is limited literature on the facilitators and barriers to engagement in digitally enabled interventions, like remote exercise programmes with sensors, among older adults with frailty. This study aimed to explore the facilitators and barriers to engagement in digital interventions among community-dwelling older adults with and without frailty. Method Community-dwelling older adults at or above 65 years old across the United Kingdoms were invited to participate in this study. Qualitative data were collected using 1:1 semi-structured interviews to understand their experiences (SETREC 6875521). Frailty was measured using PRISMA-7 where a score of greater than two was considered Frail. Purposive sampling was conducted to ensure a representative cohort was included. Interviews were audio recorded, transcribed and analysed using Braun and Clarke thematic analysis. Results Overall, 26 participants were interviewed and 13 (50%) were females. The mean age was 74.7(SD 7.67) years old, and mean duration of the interviews was 64 (SD 21.2) minutes. Six (23%) were frail on PRISMA-7. Eight themes emerged: cost, usability and functions, personal motivation, influence of immediate network, external influences, device design, perceived health benefits, and concerns about privacy and data protection. Twenty-five subthemes which emerged were categorised into facilitators and barriers to engagement. Examples of subthemes are: clear provision of health benefits was a facilitator; concerns over privacy and data protection were barriers to engagement. Participants felt there was room to do more among stakeholders like government and technology companies. Conclusions The findings highlighted various facilitators and barriers which influenced engagement with digitally enabled interventions among community-dwelling older adults with and without frailty. Wider applications of digitally enabled interventions can be informed by recommendations to overcome barriers to engagement.

Blockchain in Hand Surgery.

How does privacy protection affect the innovation performance of artificial intelligence corporations: empirical evidence from the personal information protection law

Face Recognition (FR) technology has made significant strides with the emergence of deep learning. Typically, most existing FR models are built upon Convolutional Neural Networks (CNN) and take RGB face images as the model's input. In this work, we take a closer look at existing FR paradigms from high-efficiency, security, and precision perspectives, and identify the following three problems: (i) CNN frameworks are vulnerable in capturing global facial features and modeling the correlations between local facial features. (ii) Selecting RGB face images as the model's input greatly degrades the model's inference efficiency, increasing the extra computation costs. (iii) In the real-world FR system that operates on RGB face images, the integrity of user privacy may be compromised if hackers successfully penetrate and gain access to the input of this model. To solve these three issues, we propose two novel FR frameworks, i.e., TransFace and TransFace++, which successfully explore the feasibility of applying ViTs and image bytes to FR tasks, respectively. Firstly, as revealed from our observations, we find that ViTs perform vulnerably when applied to FR scenarios with extremely large datasets. We investigate the reasons for this phenomenon and discover that the existing data augmentation approaches and hard sample mining strategies are incompatible with ViTs-based FR backbone due to the lack of tailored consideration on preserving face structural information and leveraging each local token information. To remedy these problems, we first propose a superior FR model called TransFace, which contains a patch-level data augmentation strategy named Dominant Patch Amplitude Perturbation (DPAP) and a hard sample mining strategy named Entropy-guided Hard Sample Mining (EHSM). Furthermore, to improve inference efficiency and user privacy protection, we investigate the intrinsic property of image bytes and propose a superior FR model termed TransFace++. The proposed model is trained directly on image bytes, presenting a novel approach to address the aforementioned issues. Specifically, considering the importance of local correlations in bytes, an image bytes compression strategy named Topology-based Image Bytes Compression (TIBC) is introduced to extract prominent features from the raw bytes and integrate these features with byte embeddings, effectively mitigating information loss during the bytes mapping process. Moreover, to strengthen the model's perception on geometric information encoded in image bytes, a novel cross-attention module named Structure Information-guided Cross-Attention (SICA) is designed to inject structure information into byte tokens for information interaction, significantly improving the model's generalization ability. Experiments on popular face benchmarks demonstrate the superiority of our TransFace and TransFace++.

The relationship between internet addiction and help-seeking behavior for online privacy infringement: The moderating role of online privacy perception.

CMAT: A cross-model adversarial texture for scanned document privacy protection

The integration of medical open databases with artificial intelligence (AI) technologies marks a transformative era in biomedical research and health care innovation. Over the past 25 years, initiatives like PhysioNet have revolutionized data access, fostering unprecedented levels of collaboration and accelerating medical discoveries. This rise of medical open databases presents challenges, particularly in harmonizing research enablement with patient confidentiality. In response, privacy laws such as the Health Insurance Portability and Accountability Act have been established, and privacy-enhancing technologies have been adopted to maintain this delicate balance. Privacy-enhancing technologies, including differential privacy, secure multiparty computation, and notably, federated learning (FL), have become instrumental in safeguarding personal health information. FL, in particular, represents a significant advancement by enabling the development and training of AI models on decentralized data. In Taiwan, significant strides have been made in aligning with these global data-sharing and privacy standards. We have actively promoted the sharing of medical data through the development of dynamic consent systems. These systems enable individuals to control and adjust their data-sharing preferences, ensuring transparency and continuity of consent in the ever-evolving landscape of digital health. Despite the challenges associated with privacy protections, the benefits, including improved diagnostics and treatment, are substantial. The availability of open databases has notably accelerated AI research, leading to significant advancements in medical diagnostics and treatments. As the landscape of health care research continues to evolve with open science and FL, the role of medical open databases remains crucial in shaping the future of medicine, promising enhanced patient outcomes and fostering a global research community committed to ethical integrity and privacy.

The digitization of medical records in the healthcare sector demands robust mechanisms to ensure data confidentiality, integrity, and privacy. This paper proposes an innovative multi-factor authentication (MFA) mechanism that leverages smart contracts and blockchain technology to secure the tracking of medical records. The proposed system, named Blockchain Authentication with Zero-Knowledge Proof (BAZKP), provides a tamper-proof environment for storing and accessing records while preserving users’ personally identifiable information (PII). A key novelty of BAZKP lies in storing only the character count structure of passwords rather than the actual credentials, combined with zero-knowledge proofs (ZKP) to verify identity without exposing sensitive data. This hybrid blockchain/ZKP approach addresses limitations of centralized and hardware-based solutions, reducing vulnerabilities while avoiding the cost and usability constraints of dedicated hardware systems. The system was implemented and tested on a private Ethereum testnet, with a proof-of-concept application developed using Solidity, Web3.js, and MetaMask. Performance evaluation over 100 transactions for core operations (registration, login, and password reset) demonstrated practical viability: registration incurred the highest latency (≈4500 ms) and gas consumption (≈120,000 gas), while login and reset operations were more efficient (≈4000 ms/80,000 gas and ≈3500 ms/60,000 gas, respectively). Comparative security analysis against existing MFA methods—including 2FA, hardware tokens, and biometrics—confirms that BAZKP provides superior privacy protection through decentralization and ZKP, without the cost and usability drawbacks of hardware-based solutions. Overall, this approach enhances trust in digital health systems by offering a secure, transparent, and privacy-preserving authentication framework for medical data, representing a significant advancement in digital healthcare security. Keywords: Blockchain; Multi-Factor Authentication; Smart Contracts; Zero-Knowledge Proof; Medical Record Security.

As the United States and much of the world face a resurgence of authoritarianism, the critical importance of privacy cannot be overstated. Privacy serves as a fundamental safeguard against the overreach of authoritarian governments. Authoritarian power is greatly enhanced in today’s era of pervasive surveillance and relentless data collection. We are living in the age of “surveillance capitalism.” There are vast digital dossiers about every person assembled by thousands of corporations and readily available for the government to access. The federal government and some state governments are intensifying surveillance and data collection efforts, targeting immigrants, punishing those involved in seeking or providing abortion services, and cracking down on gender-affirming healthcare. Personal data is being weaponized against critics and others who resist these efforts. In this Article, I contend that privacy protections must be significantly heightened to respond to growing threats of authoritarianism. Major regulatory interventions are necessary to prevent government surveillance from being used in inimical ways. But reforming Fourth Amendment jurisprudence and government surveillance alone will not protect against many authoritarian invasions of privacy, especially given the oligarchical character of the current strain of authoritarianism. To adequately regulate government surveillance, it is essential to also regulate surveillance capitalism. Government surveillance and surveillance capitalism are two sides of the same coin. It is impossible to protect privacy from authoritarianism without addressing consumer privacy. This Article proposes regulatory measures that should be taken to address government surveillance and surveillance capitalism—on both sides of the coin—to guard against authoritarianism.

In orthognathic surgery, accurate segmentation of the pterygopalatine and mandibular canals in maxillofacial cone beam computed tomography (CBCT) scans is crucial. It provides critical information to prevent nerve damage during surgery and significantly reduces the risk of surgical complications. However, the high cost of data collection, strict patient privacy protection, and ethical constraints have hampered the performance of existing deep learning methods for pterygopalatine and mandibular canals segmentation, limiting their practical applicability in clinical settings. To address this challenge and advance the development of pterygopalatine and mandibular canal segmentation techniques in maxillofacial CBCT scans, we carefully constructed and made publicly available a large dataset for pterygopalatine and mandibular canal segmentation in maxillofacial CBCT scans. This dataset includes 191 patient cases and comprehensively covers the key anatomical structures of the maxillary pterygopalatine canal and the mandibular canal, both of which are crucial in orthognathic surgery. Notably, this dataset is the first to include data on the maxillary pterygopalatine canal, filling a significant gap in this field. The release of this dataset will greatly accelerate the development of deep learning-based segmentation methods, provide clinicians with more accurate reconstruction tools, and ultimately improve the safety and efficiency of surgical procedures.

The digital art industry faces critical challenges in copyright protection and privacy preservation that existing solutions fail to adequately address. Traditional digital watermarking techniques are vulnerable to removal attacks and cannot prevent unauthorized content access, while current Non-Fungible Token (NFT) platforms expose transaction details and artwork content due to blockchain transparency, creating privacy risks for creators and collectors. Conventional encryption methods require decryption before any data processing, making copyright verification and feature extraction impossible in encrypted states, thus creating a fundamental security-usability trade-off. To overcome these limitations, this research proposes a network security protection system integrating homomorphic encryption with NFT copyright protection. Homomorphic encryption was selected because it uniquely enables computational operations on encrypted data without decryption, allowing copyright verification while maintaining complete data confidentiality – a capability unmatched by alternative privacy-preserving technologies. The system employs the Cheon-Kim-Kim-Song (CKKS) homomorphic encryption algorithm to construct a three-tier protection architecture consisting of an encryption layer, verification layer, and storage layer. This architecture achieves copyright verification and feature extraction of digital artworks in ciphertext state by integrating zero-knowledge proof for identity authentication and Shamir’s secret sharing for secure key management. The NFT copyright protection mechanism introduces homomorphic watermark embedding and smart contract verification, combined with proxy re-encryption to implement secure copyright transfer. A prototype system was developed and evaluated through comprehensive testing. Security performance was assessed using six metrics: privacy protection strength, copyright verification accuracy, anti-tampering capability, key security, transaction anonymity, and system resilience. Each metric was scored on a 0–100 scale based on standardized penetration testing and cryptographic attack simulations, with the comprehensive security score calculated as the weighted average of all metrics. Performance testing on 100 digital artworks across five resolutions (256×256 to 4096×4096 pixels) demonstrates that encryption time for 512×512 resolution images is kept within 15 seconds, while security testing reveals the system achieves a comprehensive security score of 94.7, representing a 60.5% improvement over traditional NFT platforms. This solution provides a practical copyright protection framework balancing security and usability for the digital art industry, with significant theoretical value and broad application prospects.

With the rapid development of Artificial Intelligence (AI) and Machine Learning (ML) technologies, federated learning models have been widely applied in various fields such as financial risk management, healthcare, and intelligent recognition. This study primarily analyzes the architectural characteristics, workflow, advantages, and disadvantages of three classical federated learning frameworks, namely horizontal federated learning, vertical federated learning, and federated transfer learning to identify the most suitable application scenarios, potential risks, and corresponding security solutions for each model. Furthermore, the paper discusses several major challenges commonly faced by different federated learning models, including issues such as regulatory compliance and cross-domain interoperability. Based on the analytical findings, it also explores potential future development directions of federated learning, such as the maturation of a trustworthy federated AI ecosystem and governance. Finally, the study concludes that implementing multi-layer privacy protection strategies, encrypted gradient exchange mechanisms and other methods can effectively reduce security risks in federated learning systems.

The rapid growth of mobile payment systems has positioned Near Field Communication (NFC) as a core enabling technology. However, conventional NFC protocols primarily emphasize transmission efficiency rather than robust authentication and privacy protection, which exposes users to threats such as eavesdropping, replay, and tracking attacks. In this study, a lightweight and privacy-preserving authentication protocol is proposed for NFC-based mobile payment services. The protocol integrates anonymous authentication, replay resistance, and tracking protection while maintaining low computational overhead suitable for resource-constrained devices. A secure offline session key generation mechanism is incorporated to enhance transaction reliability without increasing system complexity. Formal security verification using the Scyther tool (version 1.1.3) confirms resistance against major attack vectors, including impersonation, man-in-the-middle, and replay attacks. Comparative performance analysis further demonstrates that the proposed scheme achieves superior efficiency and stronger security guarantees compared with existing approaches. These results indicate that the protocol provides a practical and scalable solution for secure and privacy-aware NFC mobile payment environments.

While anonymity networks such as Tor provide invaluable privacy guarantees to society, they also enable all kinds of criminal activities. Consequently, many blameless citizens shy away from protecting their privacy using such technology for fear of being associated with criminals. To grasp the potential for alternative privacy protection for those users, we design Seldom , an anonymity network with integrated selective deanonymization that disincentivizes criminal activity. Seldom enables law enforcement agencies to selectively access otherwise anonymized identities of misbehaving users while providing technical guarantees preventing these access rights from being misused. Seldom further ensures translucency , as each access request is approved by a trustworthy consortium of impartial entities and eventually disclosed to the public (without interfering with ongoing investigations). To demonstrate Seldom ’s feasibility and applicability, we base our implementation on Tor, the most widely used anonymity network. Our evaluation indicates minimal latency, processing, and bandwidth overheads compared to Tor; Seldom ’s main costs stem from storing flow records and encrypted identities. With at most 636 TB of storage required in total to retain the encrypted identifiers of a Tor-sized network for two years, Seldom provides a practical and deployable technical solution to the inherent problem of criminal activities in anonymity networks. As such, Seldom sheds new light on the potentials and limitations when integrating selective deanonymization into anonymity networks.

Zero-Knowledge Mandates: Privacy-Preserving Delegation &amp; Spend Controls for AP2 Across Heterogeneous Rails

Sitting posture is closely related to our health. Poor sitting posture can cause various diseases and jeopardize our health. Among the current methods for detecting sitting posture, computer vision solutions suffer from privacy leakage and wearable sensor solutions suffer from inconvenience and cost of wearing. In this study, we introduce 3D-Sitpose, which leverages millimeter-wave radar to detect human sitting posture. 3D-Sitpose utilizes wireless signal transmission for non-contact detection, ensuring privacy protection and cost reduction. Firstly, we analyze the impact of variations in human sitting posture on millimeter-wave radar signals, and design sophisticated signal processing methods to refine the collected radar data, yielding clearer point cloud information for volunteers in various sitting postures. Secondly, we develop a two-channel neural network to extract fine-grained features related to volunteers from the point cloud data. Finally, we obtain coordinates for 25 human skeletal points. 3D-Sitpose can instruct users to maintain correct sitting posture based on a set of six key angles. We recruit 20 volunteers from our institute to conduct comprehensive evaluations of 3D-Sitpose. Experiments are conducted in two indoor environments to estimate sitting posture. The results reveal the mean Euclidean distance error for all skeletal point locations is 6.65 cm. This demonstrates that our method is able to estimate various sitting changes in volunteers.