While 96% of U.S. teens use the internet daily, most families face challenges in discussing privacy concerns, with parents feeling unprepared and teens being hesitant to communicate. This study explored how guided family discussions, grounded in perspective-taking theory, promoted mutual understanding and enhanced digital privacy literacy. Through a qualitative study involving 13 parent-child pairs, we identified three key communication challenges: abstract discussions about privacy, reliance on absolute statements, and a decline in teen engagement. These challenges stemmed from limited privacy literacy and a lack of adaptive communication. Our perspective-taking facilitation approach addressed these issues by transforming traditional parent-led conversations into collaborative exchanges through reflective practices and helping families view privacy as a context-dependent concept. We propose design implications for educational technology to scale the support of family privacy discussions, including tools that support perspective-taking and interfaces that highlight non-binary privacy choices.

As families face increasingly complex safety challenges in digital and physical environments, generative AI (GenAI) presents new opportunities to support household safety through multiple specialized AI agents. Through a two-phase qualitative study consisting of individual interviews and collaborative sessions with 13 parent-child dyads, we explored families' conceptualizations of GenAI and their envisioned use of AI agents in daily family life. Our findings reveal that families preferred to distribute safety-related support across multiple AI agents, each embodying a familiar caregiving role: a household manager coordinating routine tasks and mitigating risks such as digital fraud and home accidents; a private tutor providing personalized educational support, including safety education; and a family therapist offering emotional support to address sensitive safety issues such as cyberbullying and digital harassment. Families emphasized the need for agent-specific privacy boundaries, recognized generational differences in trust toward AI agents, and stressed the importance of maintaining open family communication alongside the assistance of AI agents. Based on these findings, we propose a multi-agent system design featuring four privacy-preserving principles: memory segregation, conversational consent, selective data sharing, and progressive memory management to help balance safety, privacy, and autonomy within family contexts.

The shuffle model employs a shuffler to anonymize and permute user messages, thereby enhancing privacy/utility trade-offs compared to the local model. Ideally, it assumes perfect message anonymity protection against adversaries, allowing each user to hide among a large population. However, in contexts like mobile/edge networks or in scenarios where the shuffler is curious, this assumption is frequently unrealistic. In this study, we demonstrate the vulnerability of the shuffle model to communication side-channel attacks, which substantially compromise privacy amplification via shuffling. We categorize side-channel information in the shuffle model into three types: (i) in-out information, revealing the victim user’s participation and timing, (ii) message-cardinality information, indicating the victim’s message count, and (iii) message-length information, disclosing the victim’s message length(s). Numerical results indicate these attacks increase privacy loss by 200% to 4100%, revealing secret value with probability more than 90%. After theoretically analyzing the remaining privacy amplification effects, we suggest several countermeasures and principles to alleviate degradation caused by these attacks: (a) appending padding bits to each message to counter message-length attacks, (b) maximizing query parallelization to elude in-out attacks and increase the population for privacy amplification, and (c) sending dummy messages to exchange communication costs for improved privacy amplification effects. The newly proposed paradigms and principles significantly save privacy budget in comparison to current models under attack.

The shuffle model of differential privacy provides promising privacy-utility balances in decentralized, privacy-preserving data analysis. However, the current analyses of privacy amplification via shuffling lack both tightness and generality. To address this issue, we propose the variation-ratio reduction as a comprehensive framework for privacy amplification in both single-message and multi-message shuffle protocols. It leverages two new parameterizations: the total variation bounds of local messages and the probability ratio bounds of blanket messages, to determine indistinguishability levels. Our theoretical results demonstrate that our framework provides tighter bounds, especially for local randomizers with extremal probability design, where our bounds are exactly tight. Additionally, variation-ratio reduction complements parallel composition in the shuffle model, yielding enhanced privacy accounting for popular sampling-based randomizers employed in statistical queries (e.g., range queries, marginal queries, and frequent itemset mining). Empirical findings demonstrate that our numerical amplification bounds surpass existing ones, conserving up to 30% of the budget for single-message protocols, 75% for multi-message ones, and a striking 75%-95% for parallel composition. Our bounds also result in a remarkably efficient Õ ( n ) algorithm that numerically amplifies privacy in less than 10 seconds for n = 10 8 users.

The IDC research community has a growing interest in designing AI interfaces for children with special educational needs. Nonetheless, little research has explored the research and design issues, rationale, challenges, and opportunities in this field. Therefore, we propose to host a half-day workshop to bring together researchers and practitioners from the Learning & Education, Accessibility, and Intelligent User Interfaces sub-fields to discuss and identify existing design issues, challenges, and collaboration barriers, to establish consensus on the design of a pragmatic framework, as well as explore future innovation and research opportunities. We aim to foster mutual understanding and in-depth collaboration among researchers in the IDC community.

Most students with specific learning disabilities (SLDs) have difficulties in reading and writing. The SLDs pre-screening is crucial because the golden period for therapy is before six years old. However, many students in Hong Kong receive SLDs assessments after the golden period. Also, the SLDs pre-screening is challenging, especially in a language with the logographic script but without prominent sound-script correspondence (e.g., Chinese, Japanese). To make pre-screening SLDs in Chinese more effective and efficient, we designed a new comprehensive pre-screening game for SLDs in Chinese (i.e., dyslexia, dysgraphia, and dyspraxia). Notably, we designed a Chinese morphological awareness puzzle that challenges students to recognize different words made up with the first character that is identical and the second character that is different, such as樹枝 (literally means tree branch),樹幹 (literally means tree truck),樹葉 (literally means tree leaves), and樹根 (literally means tree root). We experimented with students, which showed that our game can effectively pre-screen students with SLDs in Chinese. Our work contributes an approach to quick SLDs in Chinese pre-screening, potentially useful for other logographic languages (e.g., Japanese).

Students with specific learning disabilities (SLDs) often experience reading, writing, attention, and physical movement coordination difficulties. However, in Hong Kong, it takes years for special education needs coordinators (SENCOs) and special-ed teachers to pre-screen and diagnose students with SLDs. Therefore, many students with SLDs missed the golden time for special interventions (i.e., before six years old). In addition, although there are screening tools for students with SLDs in Chinese and Indo-European languages (e.g., English and Spanish), they did not provide a student data visualization dashboard that could help teachers speed up the pre-screening process. Therefore, we designed a new visualization dashboard for Hong Kong SENCOs and special-ed teachers to assist them in pre-screening students with SLDs. Our formative study showed that our current design met teachers’ need to quickly identify a student’s specific under-performing tasks and effectively collect evidence about how the student was affected by SLDs. Future work will further test the efficacy of our design in real life.

Students with specific learning disabilities (SLDs) often experience negative emotions when solving math problems, which they have difficulty managing. This is one reason that current math e-learning tools, which elicit these negative emotions, are not effective for these students. We designed an intelligent math e-tutoring system that aims to reduce students’ negative emotional behaviors. The system automatically detects possible negative emotional behaviors by analyzing gaze, inputs on the touchscreen, and response time. It then uses one of four intervention methods (e.g., hints or brain breaks) to prevent students from being upset. To form this design, we conducted a formative study with five teachers for students with SLDs. The teachers thought that the design of four intervention methods would help students with SLDs. Among the four intervention methods, providing brain breaks is new and particularly useful for the students. The teachers also suggested that the system should personalize the detection of negative emotional behaviors to help students who have more severe learning disabilities.

Many students with specific learning disabilities (SLDs) have difficulty learning math. To succeed in math, they need to receive personalized support from teachers. Recently, math e-learning tools that provide personalized math skills training have gained popularity. However, we know little about how well these tools help teachers personalize instruction for students with SLDs. To answer this question, we conducted semi-structured interviews with 12 teachers who taught students with SLDs in grades five to eight. We found that participants used math e-learning tools that were not designed specifically for students with SLDs. Participants had difficulty using these tools because of text-intensive user interfaces, insufficient feedback about student performance, inability to adjust difficulty levels, and problems with setup and maintenance. Participants also needed assistive technology for their students, but they had challenges in getting and using it. From our findings, we distilled design implications to help shape the design of more inclusive and effective e-learning tools.

Students with specific learning disabilities (SLD) typically struggle in their K-12 math classes, limiting the likelihood of success in STEM fields. Private tutoring is reported to be effective at helping them succeed in math, but it is not a scalable solution. While many recent e-learning tools have aimed at personalizing math support in ways that might be scalable, there remains much to be done. To better understand the gaps between current tools and the particular needs of students with SLD (and of their teachers), we conducted semi-structured interviews with 10 middle school math teachers. Our findings shed light on both the learning challenges faced by students with SLD and on the instructional challenges their teachers experience with e-learning tools. Further, we came to appreciate the importance of harnessing teacher perspectives in the design of effective e-learning tools for special students.

Phishing attacks are a major problem, as evidenced by the DNC hackings during the 2016 US presidential election, in which staff were tricked into sharing passwords by fake Google security emails, granting access to confidential information. Vulnerabilities such as these are due in part to insufficient and tiresome user training in cybersecurity. Ideally, we would have more engaging training methods that teach cybersecurity in an active and entertaining way. To address this need, we introduce the game What.Hack, which not only teaches phishing concepts but also simulates actual phishing attacks in a role-playing game to encourage the player to practice defending themselves. Our user study shows that our game design is more engaging and effective in improving performance than a standard form of training and a competing training game design (which does not simulate phishing attempts through role-playing).

Emerging smart contract systems over decentralized cryptocurrencies allow mutually distrustful parties to transact safely without trusted third parties. In the event of contractual breaches or aborts, the decentralized blockchain ensures that honest parties obtain commensurate compensation. Existing systems, however, lack transactional privacy. All transactions, including flow of money between pseudonyms and amount transacted, are exposed on the blockchain. We present Hawk, a decentralized smart contract system that does not store financial transactions in the clear on the blockchain, thus retaining transactional privacy from the public's view. A Hawk programmer can write a private smart contract in an intuitive manner without having to implement cryptography, and our compiler automatically generates an efficient cryptographic protocol where contractual parties interact with the blockchain, using cryptographic primitives such as zero-knowledge proofs. To formally define and reason about the security of our protocols, we are the first to formalize the blockchain model of cryptography. The formal modeling is of independent interest. We advocate the community to adopt such a formal model when designing applications atop decentralized blockchains.

Large scale data processing brings new challenges to the design of privacy-preserving protocols: how to meet the increasing requirements of speed and throughput of modern applications, and how to scale up smoothly when data being protected is big. Efficiency and scalability become critical criteria for privacy preserving protocols in the age of Big Data. In this paper, we present a new Private Set Intersection (PSI) protocol that is extremely efficient and highly scalable compared with existing protocols. The protocol is based on a novel approach that we call oblivious Bloom intersection. It has linear complexity and relies mostly on efficient symmetric key operations. It has high scalability due to the fact that most operations can be parallelized easily. The protocol has two versions: a basic protocol and an enhanced protocol, the security of the two variants is analyzed and proved in the semi-honest model and the malicious model respectively. A prototype of the basic protocol has been built. We report the result of performance evaluation and compare it against the two previously fastest PSI protocols. Our protocol is orders of magnitude faster than these two protocols. To compute the intersection of two million-element sets, our protocol needs only 41 seconds (80-bit security) and 339 seconds (256-bit security) on moderate hardware in parallel mode.