2026· Proceedings of the 23rd International Conference on Security and Cryptography· 0 citations· 34 references
Abstract
: As the usage of Artificial Intelligence (AI) for sensitive purposes increases, there is a growing need for privacy-aware explainable AI (XAI) tools. In this paper, we present a privacy-preserving counterfactual explanation algorithm . Our starting point is a decision-support model that is able to operate on vertically partitioned datasets, meaning that each party holds a different subset of datapoint attributes. The goal of a counterfactual algorithm is to find, given an observation , a datapoint from the (virtual) dataset that is closest to the observation but has a different label. Our algorithm fully preserves the privacy of the n datapoints belonging to the different parties by combining the strengths of homomorphic encryption and secret sharing . Through a number of experiments, we demonstrate the added value of combining multiple datasets in a realistic scenario and show that the privacy-preserving solution does not affect the accuracy. We fully implement our solution and demonstrate that it scales as to thousands of datapoints.
This SoK model federated learning as an append-only transcript of submissions, admissions, aggregation, and finalization events, and formalize verifiability as a collection of integrity claims issued by clients and the aggregator, and checked by different verifier classes.
Andrea Rizzini, Marco Esposito, Tommaso Gagliardoni et al.· Proceedings on Privacy Enhan...· 0 citations
Privacy Pass is a cryptographic scheme for issuing one-time anonymous authorization tokens, first designed as an anti-DDoS tool and an alternative to CAPTCHAs. It has attracted a lot of attention from the industry and is currently used and supported by the technological giants such as Cloudflare, Google and Apple. At the same time, Privacy Enhancing Technologies became the focus of European and non-European lawmakers (for example in the eIDAS 2.0 regulation and GDPR). Security and privacy by-design is now quite frequently a formal requirement. Privacy Pass could be used in that context as well as a lightweight solution for many application areas, e.g., for age verification. It is therefore imperative that Privacy Pass is analyzed in all possible aspects and adversarial models that are realistic, yet have not been considered during the design process. In this work, we first prove that the three most prominent variants of Privacy Pass are anamorphic. Then, we show that anamorphism of Privacy Pass makes it insecure in a model where user's device or client application is working against them (as it can be supplied by a malicious third party, the OS might be subverted or the device could be subverted). Due to anamorphism, the attacks on unlinkability remain undetectable even if an auditor is given all private keys used in the protocol, including the signer/issuer's private key. On the positive side, anamorphism can also be used to achieve a private metadata-like functionality and utilized, for example, for lawful deanonymization of malicious users, without reshaping Privacy Pass.
Mirosław Kutyłowski, Oliwer Sobolewski· Proceedings on Privacy Enhan...· 0 citations
Now, AI runs on cloud platforms, edge systems with federated settings, and in large language model (LLM) pipelines or data-sharing services, creating even wider privacy leakage paths beyond classical database disclosure. This paper offers a systematic, structured review of the literature on a curated, cost-effective reference corpus for quantifying and preventing privacy leakage in AI-enabled data ecosystems. The review ties together four strands of research that are often treated separately. Firstly, the privacy risk throughout the AI life cycle. Secondly, the measurement of the quantitative leakage. Thirdly, architectures of the privacy-preserving models, and finally, operational governance for real-world deployment. Our analysis demonstrates that state-of-the-art approaches are moving from static mechanisms based on anonymization to metric-aware protections, including information-theoretic leakage scores, cumulative differential privacy accounting, personalized privacy budgets, and benchmark-driven attack evaluation. In parallel, prevention methods are evolving beyond single homomorphic noise injection and are becoming multi-layered defenses that combine differential privacy, federated learning, weight quantization, synthetic data generation, policy-driven automation, and LLM controls. The review uncovers four itchy gaps: fractured assessment metrics, shaky privacy-utility trade-offs, flimsy integration of technological controls and compliance processes, and low cross-context validation across cloud-based computing, edge computing (data processing at or near the source), federated learning (distributed machine-learning methods), and generative AI systems. The paper concludes by outlining a unified research agenda to build AI-aware, quantifiable, and usable privacy protection stacks.
AI verification crosses a trust boundary: a verifier must learn enough to establish an authorized claim, yet the same evidence can reveal sensitive details about the model, workload, or hardware. We introduce minimal information disclosure (MID), which designs and quantifies the information content of verifier-facing evidence itself. MID measures collateral leakage with conditional mutual information: what the release reveals about the protected property after the authorized result is known. MID is general by design: it can accommodate different verification goals, protected properties, evidence sources, and deployment constraints. To demonstrate MID's practicality, we evaluate it on four physical measurements and six verification tasks spanning execution type, hardware identity, compute scale, and model identity. These experiments use three mechanism-design variables--the evidence channel, collection policy, and release transformation--but MID is not limited to these choices and can accommodate other deployable mechanisms. Across these tasks, MID produces three releases with perfect held-out verification and zero measured collateral leakage, while the remaining tasks yield explicit privacy--utility frontiers. MID also supports ZKP-certified releases: we demonstrate our proposed linear-projection mechanism using a Groth16 zk-SNARK.
The rapid growth of distributed computing paradigms, such as the Internet of Things (IoT), edge computing, cloud computing and cyber-physical systems, has made privacypreserving anomaly detection a pressing research challenge. This paper presents a systematic literature review, conducted following the PRISMA 2020 guidelines, of 50 studies published between 2018 and 2026 that combine machine-learningbased anomaly detection with privacy-enhancing technologies. We organise the literature along four axes: detection models (Isolation Forest, autoencoders, one-class SVM, graph neural networks and transformers), learning paradigms (centralized versus federated learning, FL), privacy mechanisms (homomorphic encryption, differential privacy, secure multi-party computation and zero-knowledge proofs), and integrity mechanisms based on blockchain. The reviewed applications span IoT security, healthcare, finance, industrial control, V2X networks, the metaverse and supply-chain management. Synthesising the reported evidence, the review finds that FL combined with the lightweight Isolation Forest (IF) is the approach most frequently associated with a favourable trade-off between detection quality, privacy protection and computational cost on resource-constrained edge devices, while hybrid designs that add differential privacy or homomorphic encryption offer stronger formal guarantees at a measurable cost in accuracy and latency. We critically discuss the methodological limitations of cross-study comparison, and we identify open challenges including non-IID data distributions, resistance to poisoning attacks, post-quantum cryptographic resilience and model explainability under privacy constraints. We close with future directions: adaptive privacy-budget mechanisms, federated unlearning for the right to be forgotten, and the integration of quantum-safe cryptographic primitives. All quantitative figures reported in this review are attributed to their original studies; no new experiments were performed.
Mahnoor Fatima, Ahmad Ijaz, Aoun E. Muhammad et al.· ICACNC 2026 Proceedings· 0 citations
This study explores the integration of homomorphic encryption and differential privacy techniques to enhance data privacy and security in Federated Learning (FL) systems. FL allows data to remain on local devices, eliminating the need for centralized data collection; however, sensitive information may still be leaked during model updates. To address this issue, homomorphic encryption enables computations on encrypted data, while differential privacy prevents the extraction of individual information through statistical techniques applied to model outputs. The proposed architecture was tested on the Framingham, Pima Indians Diabetes, and Bank Marketing datasets, revealing that enhanced privacy can be achieved without significantly compromising model accuracy. Furthermore, the impact of data heterogeneity among clients on model performance was analyzed, and it was concluded that strategies such as the careful selection of differential privacy parameters and training settings, along with the use of larger datasets, can improve the efficiency of FL. The findings demonstrate that privacy-preserving and high-performance artificial intelligence systems can be securely applied in sensitive domains such as healthcare and finance.
Cagdas Karatas, Hibanur Karadogan, A. Ertug et al.· 0 citations