Research studies routinely publish summary statistics such as means and standard deviations to promote transparency while protecting participant privacy. Membership inference attacks (MIAs) can exploit these statistics to determine whether a specific individual contributed to a study, posing a risk especially in biomedical and health-related settings. However, existing attacks assume the adversary holds the exact data used in the study, an assumption that rarely holds when data evolves over time. Moreover, prior work has not quantified how much of the reported accuracy stems from true individual identification rather than from group-level traits shared within disease cohorts. We investigate the robustness and interpretability of two standard attacks—the L1-distance test and the log-likelihood ratio (LLR) test—under realistic conditions where the adversary has only noisy, partial, or temporally mismatched data. We derive a theoretical lower bound on inference error that cleanly separates a statistical term governed by pool size and feature dimensionality from a signal term capturing disease-driven shifts. Empirical evaluation on cross-sectional and longitudinal miRNA datasets, validated on Fitbit activity data, confirms that both attacks tolerate substantial noise and missing features, but that real-world temporal drift degrades accuracy far more steeply than synthetic perturbations predict, and that this degradation is individual-specific. We further show that attack accuracy on disease-specific cohorts exceeds that on size-matched random pools by approximately 10%, a separation that grows almost threefold when measured by true-positive rate at 1% false-positive rate. Moreover, individuals sharing disease traits but absent from the study are frequently misclassified as members, indicating that a substantial component of reported accuracy reflects shared condition rather than individual membership.

Sanctioning blockchain addresses has become a common regulatory response to malicious activities. However, enforcement on permissionless blockchains remains challenging due to complex transaction flows and sophisticated fund-obfuscation techniques. Using cryptocurrency mixing tool Tornado Cash as a case study, we quantitatively assess the effectiveness of U.S. Office of Foreign Assets Control (OFAC) sanctions over a 957-day period, covering 6.79 million Ethereum blocks and 1.07 billion transactions. Our analysis reveals that while OFAC sanctions reduced overall Tornado Cash deposit volume by 71.03 We identify three structural limitations in current sanction enforcement practices: (i) the susceptibility of binary sanction classifications to dusting attacks; (ii) fragmented censorship by blockchain producers; and (iii) the complexity of obfuscation services exploited by users. To address these gaps, we introduce a more practical algorithm for scoring and tracking, grounded in quantitative impurity. On average, our algorithm processes Ethereum blocks within 0.07 ± 0.03 seconds and achieves 97.61

Local Differential Privacy (LDP) provides a formal guarantee of privacy that enables the collection and analysis of sensitive data without revealing any individual's data. While LDP methods have been extensively studied, there is a lack of a systematic and empirical comparison of LDP methods for descriptive statistics. In this paper, we first provide a systematization of LDP methods for descriptive statistics, comparing their properties and requirements. We demonstrate that several mean estimation methods based on sampling from a Bernoulli distribution are equivalent in the one-dimensional case and introduce methods for variance estimation. We then empirically compare methods for mean, variance, and frequency estimation. Finally, we provide recommendations for the use of LDP methods for descriptive statistics and discuss their limitations and open questions.

Can you imagine, blockchain transactions can talk! In this paper, we study how they talk and what they talk about. We focus on the input data field of Ethereum transactions, which is designed to allow external callers to interact with smart contracts. In practice, this field also enables users to embed natural language messages into transactions. Users can leverage these Input Data Messages (IDMs) for peer-to-peer communication. This means that, beyond Ethereum's well-known role as a financial infrastructure, it also serves as a decentralized communication medium. We present the first large-scale analysis of Ethereum IDMs from the genesis block to February 2024 (3134 days). We filter IDMs to extract 867,140 transactions with informative IDMs and use LLMs for language detection. We find that English (95.4 in IDMs. Interestingly, English IDMs center on security and scam warnings (24 with predominantly negative emotions, while Chinese IDMs emphasize emotional expression and social connection (44 observe that longer English IDMs often transfer high ETH values for protocol-level purposes, while longer Chinese IDMs tend to involve symbolic transfer amounts for emotional intent. Moreover, we find that the IDM participants tend to form small, loosely connected communities (59.99 findings highlight culturally and functionally divergent use cases of the IDM channel across user communities. We further examine the security relevance of IDMs in on-chain attacks. Many victims use them to appeal to attackers for fund recovery. IDMs containing negotiations or reward offers are linked to higher reply rates. We also analyze IDMs' regulatory implications. Their misuse for abuse, threats, and sexual solicitation reveals the urgent need for content moderation and regulation in decentralized systems.

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, one of the most popular attacks that are specifically designed for GNNs. 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.

Large-scale pre-trained models are increasingly adapted to downstream tasks through a new paradigm called prompt learning. In contrast to fine-tuning, prompt learning does not update the pre-trained model's parameters. Instead, it only learns an input perturbation, namely prompt, to be added to the downstream task data for predictions. Given the fast development of prompt learning, a well-generalized prompt inevitably becomes a valuable asset as significant effort and proprietary data are used to create it. This naturally raises the question of whether a prompt may leak the proprietary information of its training data. In this paper, we perform the first comprehensive privacy assessment of prompts learned by visual prompt learning through the lens of property inference and membership inference attacks. Our empirical evaluation shows that the prompts are vulnerable to both attacks. We also demonstrate that the adversary can mount a successful property inference attack with limited cost. Moreover, we show that membership inference attacks against prompts can be successful with relaxed adversarial assumptions. We further make some initial investigations on the defenses and observe that our method can mitigate the membership inference attacks with a decent utility-defense trade-off but fails to defend against property inference attacks. We hope our results can shed light on the privacy risks of the popular prompt learning paradigm. To facilitate the research in this direction, we will share our code and models with the community.

The realm of digital health is experiencing a global surge, with mobile applications extending their reach into various facets of daily life. From tracking daily eating habits and vital functions to monitoring sleep patterns and even the menstrual cycle, these apps have become ubiquitous in their pursuit of comprehensive health insights. Many of these apps collect sensitive data and promise users to protect their privacy - often through pseudonymization. We analyze the real anonymity that users can expect by this approach and report on our findings. More concretely: We introduce the notion of conditional anonymity sets derived from statistical properties of the population; We measure anonymity sets for two real-world applications and present overarching findings from 39 countries; We develop a graphical tool for people to explore their own anonymity set. One of our case studies is a popular app for tracking the menstruation cycle. Our findings for this app show that, despite their promise to protect privacy, the collected data can be used to identify users up to groups of 5 people in 97% of all the US counties, allowing the de-anonymization of the individuals. Given that the US Supreme Court recently overturned abortion rights, the possibility of determining individuals is a calamity.

Backdoor attacks represent one of the major threats to machine learning models. Various efforts have been made to mitigate backdoors. However, existing defenses have become increasingly complex and often require high computational resources or may also jeopardize models' utility. In this work, we show that fine-tuning, one of the most common and easy-to-adopt machine learning training operations, can effectively remove backdoors from machine learning models while maintaining high model utility. Extensive experiments over three machine learning paradigms show that fine-tuning and our newly proposed super-fine-tuning achieve strong defense performance. Furthermore, we coin a new term, namely backdoor sequela, to measure the changes in model vulnerabilities to other attacks before and after the backdoor has been removed. Empirical evaluation shows that, compared to other defense methods, super-fine-tuning leaves limited backdoor sequela. We hope our results can help machine learning model owners better protect their models from backdoor threats. Also, it calls for the design of more advanced attacks in order to comprehensively assess machine learning models' backdoor vulnerabilities.

Zero-knowledge proof (ZKP) mixers are one of the most widely-used blockchain privacy solutions, operating on top of smart contract-enabled blockchains. We find that ZKP mixers are tightly intertwined with the growing number of Decentralized Finance (DeFi) attacks and Blockchain Extractable Value (BEV) extractions. Through coin flow tracing, we discover that 205 blockchain attackers and 2, 595 BEV extractors leverage mixers as their source of funds, while depositing a total attack revenue of 412.87M USD. Moreover, the US OFAC sanctions against the largest ZKP mixer, Tornado.Cash, have reduced the mixer’s daily deposits by more than . Further, ZKP mixers advertise their level of privacy through a so-called anonymity set size, which similarly to k-anonymity allows a user to hide among a set of k other users. Through empirical measurements, we, however, find that these anonymity set claims are mostly inaccurate. For the most popular mixers on Ethereum (ETH) and Binance Smart Chain (BSC), we show how to reduce the anonymity set size on average by and respectively. Our empirical evidence is also the first to suggest a differing privacy-predilection of users on ETH and BSC. State-of-the-art ZKP mixers are moreover interwoven with the DeFi ecosystem by offering anonymity mining (AM) incentives, i.e., users receive monetary rewards for mixing coins. However, contrary to the claims of related work, we find that AM does not necessarily improve the quality of a mixer’s anonymity set. Our findings indicate that AM attracts privacy-ignorant users, who then do not contribute to improving the privacy of other mixer users.

The internet is a major distribution platform for web applications, but there are no effective transparency and audit mechanisms in place for the web. Due to the ephemeral nature of web applications, a client visiting a website has no guarantee that the code it receives today is the same as yesterday, or the same as other visitors receive. Despite advances in web security, it is thus challenging to audit web applications before they are rendered in the browser. We propose Accountable JS, a browser extension and opt in protocol for accountable delivery of active content on a web page. We prototype our protocol, formally model its security properties with the Tamarin Prover, and evaluate its compatibility and performance impact with case studies including WhatsApp Web, AdSense and Nimiq. Accountability is beginning to be deployed at scale, with Meta's recent announcement of Code Verify available to all 2 billion WhatsApp users, but there has been little formal analysis of such protocols. We formally model Code Verify using the Tamarin Prover and compare its properties to our Accountable JS protocol. We also compare Code Verify's and Accountable JS extension's performance impacts on WhatsApp Web.

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.

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.

The advancement of molecular profiling techniques fuels biomedical research with a deluge of data. To facilitate data sharing, the Global Alliance for Genomics and Health established the Beacon system, a search engine designed to help researchers find datasets of interest. While the current Beacon system only supports genomic data, other types of biomedical data, such as DNA methylation, are also essential for advancing our understanding in the field. In this paper, we propose the first Beacon system for DNA methylation data sharing: MBeacon. As the current genomic Beacon is vulnerable to privacy attacks, such as membership inference, and DNA methylation data is highly sensitive, we take a privacy-by-design approach to construct MBeacon. First, we demonstrate the privacy threat, by proposing a membership inference attack tailored specifically to unprotected methylation Beacons. Our experimental results show that 100 queries are sufficient to achieve a successful attack with AUC (area under the ROC curve) above 0.9. To remedy this situation, we propose a novel differential privacy mechanism, namely SVT2, which is the core component of MBeacon. Extensive experiments over multiple datasets show that SVT2 can successfully mitigate membership privacy risks without significantly harming utility. We further implement a fully functional prototype of MBeacon which we make available to the research community.

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.

Machine learning (ML) has become a core component of many real-world applications and training data is a key factor that drives current progress. This huge success has led Internet companies to deploy machine learning as a service (MLaaS). Recently, the first membership inference attack has shown that extraction of information on the training set is possible in such MLaaS settings, which has severe security and privacy implications. However, the early demonstrations of the feasibility of such attacks have many assumptions on the adversary, such as using multiple so-called shadow models, knowledge of the target model structure, and having a dataset from the same distribution as the target model's training data. We relax all these key assumptions, thereby showing that such attacks are very broadly applicable at low cost and thereby pose a more severe risk than previously thought. We present the most comprehensive study so far on this emerging and developing threat using eight diverse datasets which show the viability of the proposed attacks across domains. In addition, we propose the first effective defense mechanisms against such broader class of membership inference attacks that maintain a high level of utility of the ML model.

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%.

Privacy-Enhancing Technologies (PETs) wie Differential Privacy und anderen Anonymisierungsverfahren kommen im Hinblick auf hochsensible Gesundheitsdaten besondere Bedeutung zu. Der vorliegende Beitrag zeigt neue Datenschutzrisiken bei epigenetischen Daten auf, entwickelt und analysiert geeignete Gegenmaßnahmen durch PETs und diskutiert die rechtliche Bewertung von deren Einsatz in der medizinischen Forschung.

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 decreasing costs of molecular profiling have fueled the biomedical research community with a plethora of new types of biomedical data, allowing for a breakthrough towards a more precise and personalized medicine. However, the release of these intrinsically highly sensitive, interdependent data poses a new severe privacy threat. So far, the security community has mostly focused on privacy risks arising from genomic data. However, the manifold privacy risks stemming from other types of biomedical data – and epigenetic data in particular – have been largely overlooked. In this thesis, we provide means to quantify and protect the privacy of individuals’ biomedical data. Besides the genome, we specifically focus on two of the most important epigenetic elements influencing human health: microRNAs and DNA methylation. We quantify the privacy for multiple realistic attack scenarios, namely, (1) linkability attacks along the temporal dimension, between different types of data, and between related individuals, (2) membership attacks, and (3) inference attacks. Our results underline that the privacy risks inherent to biomedical data have to be taken seriously. Moreover, we present and evaluate solutions to preserve the privacy of individuals. Our mitigation techniques stretch from the differentially private release of epigenetic data, considering its utility, up to cryptographic constructions to securely, and privately evaluate a random forest on a patient’s data.

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.