Second-hand electronic devices are increasingly being sold online. Although more affordable and more environment-friendly than new products, second-hand devices, in particular those with storage capabilities, create security and privacy threats (e.g., malware or confidential data still stored on the device, aka remnant data). Previous work studied this issue from a technical point of view or only from the perspective of the sellers of the devices, but the perspective of the buyers has been largely overlooked. In this paper, we fill this gap and take a multi-disciplinary approach, focusing on the case of Switzerland. First, we conduct a brief legal analysis of the rights and obligations related to second-hand storage devices. Second, in order to understand the buyers' practices related to these devices and their beliefs about their legal rights and obligations, we deploy a survey in collaboration with a major online platform for transactions of second-hand goods. We demonstrate that the risks highlighted in prior research might not materialize, as many buyers do not inspect the content of the bought devices (e.g., they format it directly). We also found that none of the buyers uses forensic techniques. We identified that the buyers' decisions about remnant data depend on the type of data. For instance, for data with illegal content, they would keep the data to report it to the authorities, whereas for sensitive personal data they would either delete the data or contact the sellers. We identified several discrepancies between the actual legal rights/obligations and users' beliefs.

A graph neural network (GNN) is a type of neural network that is specifically designed to process graph-structured data. Typically, GNNs can be implemented in two settings, including the transductive setting and the inductive setting. In the transductive setting, the trained model can only predict the labels of nodes that were observed at the training time. In the inductive setting, the trained model can be generalized to new nodes/graphs. Due to its flexibility, the inductive setting is the most popular GNN setting at the moment. Previous work has shown that transductive GNNs are vulnerable to a series of privacy attacks. However, a comprehensive privacy analysis of inductive GNN models is still missing. This paper fills the gap by conducting a systematic privacy analysis of inductive GNNs through the lens of link stealing attacks. We propose two types of link stealing attacks, i.e., posterior-only attacks and combined attacks. We define threat models of the posterior-only attacks with respect to node topology and the combined attacks by considering combinations of posteriors, node attributes, and graph features. Extensive evaluation on six real-world datasets demonstrates that inductive GNNs leak rich information that enables link stealing attacks with advantageous properties. Even attacks with no knowledge about graph structures can be effective. We also show that our attacks are robust to different node similarities and different graph features. As a counterpart, we investigate two possible defenses and discover they are ineffective against our attacks, which calls for more effective defenses.

Wearable activity trackers (WATs) have recently gained worldwide popularity, with over a billion devices collecting a range of personal data. To receive additional services, users commonly share this data with third-party applications (TPAs). However, this practice poses potential privacy risks. Privacy-enhancing technologies have been developed to address these concerns, but they often lack user-centered design, and therefore, are less likely to be directly related to users' concerns and to be widely adopted. This study takes a participatory design approach involving N=26 experienced WAT users who share data with TPAs. Through a series of design sessions, participants conceptualized 19 solutions, from which we identified seven different design features. We further analyze and discuss how these features can be combined to assist users in managing their data sharing with TPAs and, therefore, enhancing their privacy. Finally, we selected the three most promising features, namely partial sharing, reminder, and revocation assistance, and conducted an online survey with N=201 WAT users to better understand the potential effectiveness and usability of these features. This work makes an important contribution by offering user-centered solutions and valuable insights for integrating privacy-enhancing technologies into the WAT ecosystem.

Internet users possess accounts on dozens of online services where they are often identified by one of their e-mail addresses. They often use the same address on multiple services and for communicating with their contacts. In this paper, we investigate attacks that enable an adversary (e.g., company, friend) to determine (stealthily or not) whether an individual, identified by their e-mail address, has an account on certain services (i.e., an account enumeration attack ). Such attacks on account privacy have serious implications as information about one’s accounts can be used to (1) profile them and (2) improve the effectiveness of phishing. We take a multifaceted approach and study these attacks through a combination of experiments (63 services), surveys (318 respondents), and focus groups (13 participants). We demonstrate the high vulnerability of popular services (93.7%) and the concerns of users about their account privacy, as well as their increased susceptibility to phishing e-mails that impersonate services on which they have an account. We also provide findings on the challenges in implementing countermeasures for service providers and on users’ ideas for enhancing their account privacy. Finally, our interaction with national data protection authorities led to the inclusion of recommendations in their developers’ guide.

Recently, the newly emerged multimodal models, which leverage both visual and linguistic modalities to train powerful encoders, have gained increasing attention. However, learning from a large-scale unlabeled dataset also exposes the model to the risk of potential poisoning attacks, whereby the adversary aims to perturb the model’s training data to trigger malicious behaviors in it. In contrast to previous work, only poisoning visual modality, in this work, we take the first step to studying poisoning attacks against multimodal models in both visual and linguistic modalities. Specially, we focus on answering two questions: (1) Is the linguistic modality also vulnerable to poisoning attacks? and (2) Which modality is most vulnerable? To answer the two questions, we propose three types of poisoning attacks against multimodal models. Extensive evaluations on different datasets and model architectures show that all three attacks can achieve significant attack performance while maintaining model utility in both visual and linguistic modalities. Furthermore, we observe that the poisoning effect differs between different modalities. To mitigate the attacks, we propose both pre-training and post-training defenses. We empirically show that both defenses can significantly reduce the attack performance while preserving the model’s utility. Our code is available at https://github.com/zqypku/mm_poison/.

The number of users of wearable activity trackers (WATs) has rapidly increased over the last decade. Although these devices enable their users to monitor their activities and health, they also raise new security and privacy concerns, given the sensitive data (e.g., steps, heart rate) they collect and the information that can be inferred from this data (e.g., diseases). In addition to sharing with the service providers (e.g., Fitbit), WAT users can share their fitness data with third-party applications (TPAs) and individuals. Understanding how and with whom users share their fitness data and what kind of approaches they take to preserve their privacy are key to assessing the underlying privacy risks and to further designing appropriate privacy-enhancing techniques. In this work, we perform, through a large-scale survey of N=628 WAT users, the first quantitative and qualitative analysis of users' awareness, understanding, attitudes, and behaviors toward fitness-data sharing with TPAs and individuals. By asking these users to draw their thoughts, we explore, in particular, users' practices and actual behaviors toward fitness-data sharing and their mental models. Our empirical results show that about half of WAT users underestimate the number of TPAs to which they have granted access to their data, and 63% share data with at least one TPA that they do not actively use (anymore). Furthermore, 29% of the users do not revoke TPA access to their data because they forget they gave access to it in the first place, and 8% were not even aware they could revoke access to their data. Finally, their mental models, as well as some of their answers, demonstrate substantial gaps in their understanding of the data-sharing process. Importantly, 67% of the respondents think that TPAs cannot access the fitness data that was collected before they granted access to it, whereas TPAs actually can do this.

Machine unlearning is a process of removing the impact of some training data from the machine learning (ML) models upon receiving removal requests. While straightforward and legitimate, retraining the ML model from scratch incurs a high computational overhead. To address this issue, a number of approximate algorithms have been proposed in the domain of image and text data, among which SISA is the state-of-the-art solution. It randomly partitions the training set into multiple shards and trains a constituent model for each shard. However, directly applying SISA to the graph data can severely damage the graph structural information, and thereby the resulting ML model utility. In this paper, we propose GraphEraser, a novel machine unlearning framework tailored to graph data. Its contributions include two novel graph partition algorithms and a learning-based aggregation method. We conduct extensive experiments on five real-world graph datasets to illustrate the unlearning efficiency and model utility of GraphEraser. It achieves 2.06x (small dataset) to 35.94x (large dataset) unlearning time improvement. On the other hand, GraphEraser achieves up to 62.5% higher F1 score and our proposed learning-based aggregation method achieves up to 112% higher F1 score. https://github.com/MinChen00/Graph-Unlearning.

IoT devices are increasingly present, both in the industry and in consumer markets, but their security remains weak, which leads to an unprecedented number of attacks against them. In order to reduce the attack surface, one approach is to analyze the binary code of these devices to early detect whether they contain potential security vulnerabilities. More specifically, knowing some vulnerable function, we can determine whether the firmware of an IoT device contains some security flaw by searching for this function. However, searching for similar vulnerable functions is in general challenging due to the fact that the source code is often not openly available and that it can be compiled for different architectures, using different compilers and compilation settings. In order to handle these varying settings, we can compare the similarity between the graph embeddings derived from the binary functions. In this paper, inspired by the recent advances in deep learning, we propose a new method – GESS (graph embeddings for similarity search) – to derive graph embeddings, and we compare it with various state-of-the-art methods. Our empirical evaluation shows that GESS reaches an AUC of 0.979, thereby outperforming the best known approach. Furthermore, for a fixed low false positive rate, GESS provides a true positive rate (or recall) about 36% higher than the best previous approach. Finally, for a large search space, GESS provides a recall between 50% and 60% higher than the best previous approach.

Direct-to-consumer genetic testing services are gaining momentum: As of today, companies such as 23andMe or AncestryDNA have already attracted 26 million customers. These services raise privacy concerns, exacerbated by the fact that their customers can then share their genomic data on platforms such as GEDmatch. Notwithstanding their right to learn about their genetic background or to share their genomic data, it is paramount that individuals realize that such a behavior damages their relatives' privacy (i.e., kin genomic privacy). In this paper, we present KGP Meter, a new online tool that provides means for raising awareness in the general public about the privacy risks of genomic data sharing. Our tool features various properties that makes it highly interactive, privacy-preserving (i.e., not requiring access to the actual genomic data), and user-friendly. It explores possible configurations in an optimized way and combines well-established graphical models with an entropy-based metric to compute kin genomic privacy scores. Our experiments show that KGP Meter is very responsive. We design and implement an interface that enables users to draw their family trees and indicate which of their relatives' genomes are known, and that communicates the resulting privacy scores to the users. We then analyze the usage of the tool and survey users to better understand users' perceptions towards these risks and evaluate our tool. We observe that most of them find the privacy score worrisome, and that the large majority of them find KGP Meter useful.

The right to be forgotten states that a data owner has the right to erase their data from an entity storing it. In the context of machine learning (ML), the right to be forgotten requires an ML model owner to remove the data owner's data from the training set used to build the ML model, a process known asmachine unlearning. While originally designed to protect the privacy of the data owner, we argue that machine unlearning may leave some imprint of the data in the ML model and thus create unintended privacy risks. In this paper, we perform the first study on investigating the unintended information leakage caused by machine unlearning. We propose a novel membership inference attack that leverages the different outputs of an ML model's two versions to infer whether a target sample is part of the training set of the original model but out of the training set of the corresponding unlearned model. Our experiments demonstrate that the proposed membership inference attack achieves strong performance. More importantly, we show that our attack in multiple cases outperforms the classical membership inference attack on the original ML model, which indicates that machine unlearning can have counterproductive effects on privacy. We notice that the privacy degradation is especially significant for well-generalized ML models where classical membership inference does not perform well. We further investigate four mechanisms to mitigate the newly discovered privacy risks and show that releasing the predicted label only, temperature scaling, and differential privacy are effective. We believe that our results can help improve privacy protection in practical implementations of machine unlearning. \footnoteOur code is available at \urlhttps://github.com/MinChen00/UnlearningLeaks.

Biomedical data sharing is one of the key elements fostering the advancement of biomedical research but poses severe risks towards the privacy of individuals contributing their data, as already demonstrated for genomic data. In this paper, we study whether and to which extent DNA methylation data, one of the most important epigenetic elements regulating human health, is prone to membership inference attacks, a critical type of attack that reveals an individual's participation in a given database. We design and evaluate three different attacks exploiting published summary statistics, among which one is based on machine learning and another is exploiting the dependencies between genome and methylation data. Our extensive evaluation on six datasets containing a diverse set of tissues and diseases collected from more than 1,300 individuals in total shows that such membership inference attacks are effective, even when the target's methylation profile is not accessible. It further shows that the machine-learning approach outperforms the statistical attacks, and that learned models are transferable across different datasets.

Web developers can (and do) include subresources such as scripts, stylesheets and images in their webpages. Such subresources might be stored on content delivery networks (CDNs). This practice creates security and privacy risks, should a subresource be corrupted. The subresource integrity (SRI) recommendation, released in mid-2016 by the W3C, enables developers to include digests in their webpages in order for web browsers to verify the integrity of subresources before loading them. In this paper, we conduct the first large-scale longitudinal study of the use of SRI on the Web by analyzing massive crawls (≈ 3B URLs) of the Web over the last 3.5 years. Our results show that the adoption of SRI is modest (≈), but grows at an increasing rate and is highly influenced by the practices of popular library developers (e.g., Bootstrap) and CDN operators (e.g., jsDelivr). We complement our analysis about SRI with a survey of web developers (N=): It shows that a substantial proportion of developers know SRI and understand its basic functioning, but most of them ignore important aspects of the recommendation. The results of the survey also show that the integration of SRI by developers is mostly manual – hence not scalable and error prone. This calls for a better integration of SRI in build tools.

App stores provide access to millions of different programs that users can download on their computers. Developers can also make their programs available for download on their websites and host the program files either directly on their website or on third-party platforms, such as mirrors. In the latter case, as users download the software without any vetting from the developers, they should take the necessary precautions to ensure that it is authentic. One way to accomplish this is to check that the published file’s integrity verification code—the checksum—matches that (if provided) of the downloaded file. To date, however, there is little evidence to suggest that such a process is effective. Even worse, very few usability studies about it exist. In this article, we provide the first comprehensive study that assesses the usability and effectiveness of the manual checksum verification process. First, by means of an in-situ experiment with 40 participants and eye-tracking technology, we show that the process is cumbersome and error-prone. Second, after a 4-month-long in-the-wild experiment with 134 participants, we demonstrate how our proposed solution—a Chrome extension that verifies checksums automatically—significantly reduces human errors, improves coverage, and has only limited impact on usability. It also confirms that, sadly, only a tiny minority of websites that link to executable files in our sample provide checksums (0.01%), which is a strong call to action for web standards bodies, service providers, and content creators to increase the use of file integrity verification on their properties.

Abstract Most popular location-based social networks, such as Facebook and Foursquare, let their (mobile) users post location and co-location (involving other users) information. Such posts bring social benefits to the users who post them but also to their friends who view them. Yet, they also represent a severe threat to the users’ privacy, as co-location information introduces interdependences between users. We propose the first game-theoretic framework for analyzing the strategic behaviors, in terms of information sharing, of users of OSNs. To design parametric utility functions that are representative of the users’ actual preferences, we also conduct a survey of 250 Facebook users and use conjoint analysis to quantify the users’ benefits o f sharing vs. viewing (co)-location information and their preference for privacy vs. benefits. Our survey findings expose the fact that, among the users, there is a large variation, in terms of these preferences. We extensively evaluate our framework through data-driven numerical simulations. We study how users’ individual preferences influence each other’s decisions, we identify several factors that significantly affect these decisions (among which, the mobility data of the users), and we determine situations where dangerous patterns can emerge (e.g., a vicious circle of sharing, or an incentive to over-share) – even when the users share similar preferences.

Abstract The decreasing costs of molecular profiling have fueled the biomedical research community with a plethora of new types of biomedical data, enabling a breakthrough towards more precise and personalized medicine. Naturally, the increasing availability of data also enables physicians to compare patients’ data and treatments easily and to find similar patients in order to propose the optimal therapy. Such similar patient queries (SPQs) are of utmost importance to medical practice and will be relied upon in future health information exchange systems. While privacy-preserving solutions have been previously studied, those are limited to genomic data, ignoring the different newly available types of biomedical data. In this paper, we propose new cryptographic techniques for finding similar patients in a privacy-preserving manner with various types of biomedical data, including genomic, epigenomic and transcriptomic data as well as their combination. We design protocols for two of the most common similarity metrics in biomedicine: the Euclidean distance and Pearson correlation coefficient. Moreover, unlike previous approaches, we account for the fact that certain locations contribute differently to a given disease or phenotype by allowing to limit the query to the relevant locations and to assign them different weights. Our protocols are specifically designed to be highly efficient in terms of communication and bandwidth, requiring only one or two rounds of communication and thus enabling scalable parallel queries. We rigorously prove our protocols to be secure based on cryptographic games and instantiate our technique with three of the most important types of biomedical data – namely DNA, microRNA expression, and DNA methylation. Our experimental results show that our protocols can compute a similarity query over a typical number of positions against a database of 1,000 patients in a few seconds. Finally, we propose and formalize strategies to mitigate the threat of malicious users or hospitals.

Internet users can download software for their computers from app stores (e.g., Mac App Store and Windows Store) or from other sources, such as the developers' websites. Most Internet users in the US rely on the latter, according to our representative study, which makes them directly responsible for the content they download. To enable users to detect if the downloaded files have been corrupted, developers can publish a checksum together with the link to the program file; users can then manually verify that the checksum matches the one they obtain from the downloaded file. In this paper, we assess the prevalence of such behavior among the general Internet population in the US (N=2,000), and we develop easy-to-use tools for users and developers to automate both the process of checksum verification and generation. Specifically, we propose an extension to the recent W3C specification for sub-resource integrity in order to provide integrity protection for download links. Also, we develop an extension for the popular Chrome browser that computes and verifies checksums of downloaded files automatically, and an extension for the WordPress CMS that developers can use to easily attach checksums to their remote content. Our in situ experiments with 40participants demonstrate the usability and effectiveness issues of checksums verification, and shows user desirability for our extension.

The decreasing costs of molecular profiling has fueled the biomedical research community with a plethora of new types of biomedical data, enabling a breakthrough towards a more precise and personalized medicine. However, the release of these intrinsically highly sensitive data poses a new severe privacy threat. While biomedical data is largely associated with our health, there also exist various correlations between different types of biomedical data, along the temporal dimension, and also in-between family members. However, so far, the security community has focused on privacy risks stemming from genomic data, largely overlooking the manifold interdependencies between other biomedical data. In this paper, we present a generic framework for quantifying the privacy risks in biomedical data taking into account the various interdependencies between data (i) of different types, (ii) from different individuals, and (iii) at different time. To this end, we rely on a Bayesian network model that allows us to take all aforementioned dependencies into account and run exact probabilistic inference attacks very efficiently. Furthermore, we introduce a generic algorithm for building the Bayesian network, which encompasses expert knowledge for known dependencies, such as genetic inheritance laws, and learns previously unknown dependencies from the data. Then, we conduct a thorough inference risk evaluation with a very rich dataset containing genomic and epigenomic data of mothers and children over multiple years. Besides effective probabilistic inference, we further demonstrate that our Bayesian network model can also serve as a building block for other attacks. We show that, with our framework, an adversary can efficiently identify the parent-child relationships based on methylation data with a success rate of 95%.

Hashtag has emerged as a widely used concept of popular culture and campaigns, but its implications on people»s privacy have not been investigated so far. In this paper, we present the first systematic analysis of privacy issues induced by hashtags. We concentrate in particular on location, which is recognized as one of the key privacy concerns in the Internet era. By relying on a random forest model, we show that we can infer a user»s precise location from hashtags with accuracy of 70% to 76%, depending on the city. To remedy this situation, we introduce a system called Tagvisor that systematically suggests alternative hashtags if the user-selected ones constitute a threat to location privacy. Tagvisor realizes this by means of three conceptually different obfuscation techniques and a semantics-based metric for measuring the consequent utility loss. Our findings show that obfuscating as little as two hashtags already provides a near-optimal trade-off between privacy and utility in our dataset. This in particular renders Tagvisor highly time-efficient, and thus, practical in real-world settings.

The dramatically decreasing costs of DNA sequencing have triggered more than a million humans to have their genotypes sequenced. Moreover, these individuals increasingly make their genomic data publicly available, thereby creating privacy threats for themselves and their relatives because of their DNA similarities. More generally, an entity that gains access to a significant fraction of sequenced genotypes might be able to infer even the genomes of unsequenced individuals. In this paper, we propose a simulation-based model for quantifying the impact of continuously sequencing and publicizing personal genomic data on a population's genomic privacy. Our simulation probabilistically models data sharing and takes into account events such as migration and interracial mating. We exemplarily instantiate our simulation with a sample population of 1,000 individuals and evaluate the privacy under multiple settings over 6,000 genomic variants and a subset of phenotype-related variants. Our findings demonstrate that an increasing sharing rate in the future entails a substantial negative effect on the privacy of all older generations. Moreover, we find that mixed populations face a less severe erosion of privacy over time than more homogeneous populations. Finally, we demonstrate that genomic-data sharing can be much more detrimental for the privacy of the phenotype-related variants.

The development of positioning technologies has resulted in an increasing amount of mobility data being available. While bringing a lot of convenience to people's life, such availability also raises serious concerns about privacy. In this paper, we concentrate on one of the most sensitive information that can be inferred from mobility data, namely social relationships. We propose a novel social relation inference attack that relies on an advanced feature learning technique to automatically summarize users' mobility features. Compared to existing approaches, our attack is able to predict any two individuals' social relation, and it does not require the adversary to have any prior knowledge on existing social relations. These advantages significantly increase the applicability of our attack and the scope of the privacy assessment. Extensive experiments conducted on a large dataset demonstrate that our inference attack is effective, and achieves between 13% to 20% improvement over the best state-of-the-art scheme. We propose three defense mechanisms -- hiding, replacement and generalization -- and evaluate their effectiveness for mitigating the social link privacy risks stemming from mobility data sharing. Our experimental results show that both hiding and replacement mechanisms outperform generalization. Moreover, hiding and replacement achieve a comparable trade-off between utility and privacy, the former preserving better utility and the latter providing better privacy.

In a world where traditional notions of privacy are increasingly challenged by the myriad companies that collect and analyze our data, it is important that decision-making entities are held accountable for unfair treatments arising from irresponsible data usage. Unfortunately, a lack of appropriate methodologies and tools means that even identifying unfair or discriminatory effects can be a challenge in practice. We introduce the unwarranted associations (UA) framework, a principled methodology for the discovery of unfair, discriminatory, or offensive user treatment in data-driven applications. The UA framework unifies and rationalizes a number of prior attempts at formalizing algorithmic fairness. It uniquely combines multiple investigative primitives and fairness metrics with broad applicability, granular exploration of unfair treatment in user subgroups, and incorporation of natural notions of utility that may account for observed disparities. We instantiate the UA framework in FairTest, the first comprehensive tool that helps developers check data-driven applications for unfair user treatment. It enables scalable and statistically rigorous investigation of associations between application outcomes (such as prices or premiums) and sensitive user attributes (such as race or gender). Furthermore, FairTest provides debugging capabilities that let programmers rule out potential confounders for observed unfair effects. We report on use of FairTest to investigate and in some cases address disparate impact, offensive labeling, and uneven rates of algorithmic error in four data-driven applications. As examples, our results reveal subtle biases against older populations in the distribution of error in a predictive health application and offensive racial labeling in an image tagger.

Since the first whole-genome sequencing, the biomedical research community has made significant steps towards a more precise, predictive and personalized medicine. Genomic data is nowadays widely considered privacy-sensitive and consequently protected by strict regulations and released only after careful consideration. Various additional types of biomedical data, however, are not shielded by any dedicated legal means and consequently disseminated much less thoughtfully. This in particular holds true for DNA methylation data as one of the most important and well-understood epigenetic element influencing human health. In this paper, we show that, in contrast to the aforementioned belief, releasing one's DNA methylation data causes privacy issues akin to releasing one's actual genome. We show that already a small subset of methylation regions influenced by genomic variants are sufficient to infer parts of someone's genome, and to further map this DNA methylation profile to the corresponding genome. Notably, we show that such re-identification is possible with 97.5% accuracy, relying on a dataset of more than 2500 genomes, and that we can reject all wrongly matched genomes using an appropriate statistical test. We provide means for countering this threat by proposing a novel cryptographic scheme for privately classifying tumors that enables a privacy-respecting medical diagnosis in a common clinical setting. The scheme relies on a combination of random forests and homomorphic encryption, and it is proven secure in the honest-but-curious model. We evaluate this scheme on real DNA methylation data, and show that we can keep the computational overhead to acceptable values for our application scenario.

The rapid progress in human-genome sequencing is leading to a high availability of genomic data. These data is notoriously very sensitive and stable in time, and highly correlated among relatives. In this article, we study the implications of these familial correlations on kin genomic privacy. We formalize the problem and detail efficient reconstruction attacks based on graphical models and belief propagation. With our approach, an attacker can infer the genomes of the relatives of an individual whose genome or phenotype are observed by notably relying on Mendel’s Laws, statistical relationships between the genomic variants, and between the genome and the phenotype. We evaluate the effect of these dependencies on privacy with respect to the amount of observed variants and the relatives sharing them. We also study how the algorithmic performance evolves when we take these various relationships into account. Furthermore, to quantify the level of genomic privacy as a result of the proposed inference attack, we discuss possible definitions of genomic privacy metrics, and compare their values and evolution. Genomic data reveals Mendelian disorders and the likelihood of developing severe diseases, such as Alzheimer’s. We also introduce the quantification of health privacy , specifically, the measure of how well the predisposition to a disease is concealed from an attacker. We evaluate our approach on actual genomic data from a pedigree and show the threat extent by combining data gathered from a genome-sharing website as well as an online social network.

Co-location information about users is increasingly available online. For instance, mobile users more and more frequently report their co-locations with other users in the messages and in the pictures they post on social networking websites by tagging the names of the friends they are with. The users' IP addresses also constitute a source of co-location information. Combined with (possibly obfuscated) location information, such co-locations can be used to improve the inference of the users' locations, thus further threatening their location privacy: As co-location information is taken into account, not only a user's reported locations and mobility patterns can be used to localize her, but also those of her friends (and the friends of their friends and so on). In this paper, we study this problem by quantifying the effect of co-location information on location privacy, considering an adversary such as a social network operator that has access to such information. We formalize the problem and derive an optimal inference algorithm that incorporates such co-location information, yet at the cost of high complexity. We propose some approximate inference algorithms, including a solution that relies on the belief propagation algorithm executed on a general Bayesian network model, and we extensively evaluate their performance. Our experimental results show that, even in the case where the adversary considers co-locations of the targeted user with a single friend, the median location privacy of the user is decreased by up to 62 percent in a typical setting. We also study the effect of the different parameters (e.g., the settings of the location-privacy protection mechanisms) in different scenarios.

The continuous decrease in cost of molecular profiling tests is revolutionizing medical research and practice, but it also raises new privacy concerns. One of the first attacks against privacy of biological data, proposed by Homer et al. in 2008, showed that, by knowing parts of the genome of a given individual and summary statistics of a genome-based study, it is possible to detect if this individual participated in the study. Since then, a lot of work has been carried out to further study the theoretical limits and to counter the genome-based membership inference attack. However, genomic data are by no means the only or the most influential biological data threatening personal privacy. For instance, whereas the genome informs us about the risk of developing some diseases in the future, epigenetic biomarkers, such as microRNAs, are directly and deterministically affected by our health condition including most common severe diseases. In this paper, we show that the membership inference attack also threatens the privacy of individuals contributing their microRNA expressions to scientific studies. Our results on real and public microRNA expression data demonstrate that disease-specific datasets are especially prone to membership detection, offering a true-positive rate of up to 77% at a false-negative rate of less than 1%. We present two attacks: one relying on the L_1 distance and the other based on the likelihood-ratio test. We show that the likelihood-ratio test provides the highest adversarial success and we derive a theoretical limit on this success. In order to mitigate the membership inference, we propose and evaluate both a differentially private mechanism and a hiding mechanism. We also consider two types of adversarial prior knowledge for the differentially private mechanism and show that, for relatively large datasets, this mechanism can protect the privacy of participants in miRNA-based studies against strong adversaries without degrading the data utility too much. Based on our findings and given the current number of miRNAs, we recommend to only release summary statistics of datasets containing at least a couple of hundred individuals.

Abstract People increasingly have their genomes sequenced and some of them share their genomic data online. They do so for various purposes, including to find relatives and to help advance genomic research. An individual’s genome carries very sensitive, private information such as its owner’s susceptibility to diseases, which could be used for discrimination. Therefore, genomic databases are often anonymized. However, an individual’s genotype is also linked to visible phenotypic traits, such as eye or hair color, which can be used to re-identify users in anonymized public genomic databases, thus raising severe privacy issues. For instance, an adversary can identify a target’s genome using known her phenotypic traits and subsequently infer her susceptibility to Alzheimer’s disease. In this paper, we quantify, based on various phenotypic traits, the extent of this threat in several scenarios by implementing de-anonymization attacks on a genomic database of OpenSNP users sequenced by 23andMe. Our experimental results show that the proportion of correct matches reaches 23% with a supervised approach in a database of 50 participants. Our approach outperforms the baseline by a factor of four, in terms of the proportion of correct matches, in most scenarios. We also evaluate the adversary’s ability to predict individuals’ predisposition to Alzheimer’s disease, and we observe that the inference error can be halved compared to the baseline. We also analyze the effect of the number of known phenotypic traits on the success rate of the attack. As progress is made in genomic research, especially for genotype-phenotype associations, the threat presented in this paper will become more serious.

The rapid progress in human-genome sequencing is leading to a high availability of genomic data. This data is notoriously very sensitive and stable in time. It is also highly correlated among relatives. A growing number of genomes are becoming accessible online (e.g., because of leakage, or after their posting on genome-sharing websites). What are then the implications for kin genomic privacy? We formalize the problem and detail an efficient reconstruction attack based on graphical models and belief propagation. With this approach, an attacker can infer the genomes of the relatives of an individual whose genome is observed, relying notably on Mendel's Laws and statistical relationships between the nucleotides (on the DNA sequence). Then, to quantify the level of genomic privacy as a result of the proposed inference attack, we discuss possible definitions of genomic privacy metrics. Genomic data reveals Mendelian diseases and the likelihood of developing degenerative diseases such as Alzheimer's. We also introduce the quantification of health privacy, specifically the measure of how well the predisposition to a disease is concealed from an attacker. We evaluate our approach on actual genomic data from a pedigree and show the threat extent by combining data gathered from a genome-sharing website and from an online social network.