Qi An, Frank Jiang, Chengzu Dong, Shantanu Pal · 7 authors
The rapid expansion of electric vehicle (EV) infrastructure necessitates advanced solutions for secure and private authentication at EV charging stations. This research introduces a blockchain-based framework enhanced with self-sovereign identity (SSI) features, targeting the improvement of privacy and security in cyber marketplaces for EVs. The inclusion of SSI enables users to maintain full control over their digital identities, a critical advancement for authentication processes at EV charging stations. This system effectively addresses the growing privacy and security challenges within the expanding EV infrastructure. By integrating Zero-knowledge proof with self-sovereign identity, the framework not only ensures robust security but also preserves user privacy by enabling users to prove their identity without exposing sensitive personal information. We propose an efficient and user-friendly solution, showcasing its potential as a pioneering innovation in the field of EV charging infrastructure.
Collision-resistant cryptographic hash functions (CRHs) are crucial for security, particularly for message authentication in Zero-knowledge Proof (ZKP) applications. However, traditional CRHs like SHA-2 or SHA-3, while optimized for CPUs, generate large circuits, rendering them inefficient in the ZK domain. Conversely, ZK-friendly hashes are designed for circuit efficiency but struggle on conventional hardware, often orders of magnitude slower than standard hashes due to their reliance on expensive finite field arithmetic. To bridge this performance gap, we present HashEmAll, a novel collection of FPGA-based realizations for three prominent ZK-friendly hashes: Griffin, Rescue-Prime, and Reinforced Concrete. Each offers distinct optimization profiles, with both area-optimized and latency-optimized variants available, allowing users to tailor hardware selection to specific application constraints regarding resource utilization and performance. Our extensive evaluation shows that latency-optimized HashEmAll designs outperform CPU implementations by at least $10 \times$, with the leading design achieving a $23 \times$ speedup. These gains are coupled with lower power consumption and compatibility with accessible FPGAs. Importantly, the highly parallel and pipelined architecture of HashEmAll enables significantly better practical scaling than CPU-based approaches towards building real-world ZKP applications, such as data commitments with Merkle Trees, by mitigating the hashing bottleneck for large trees. This highlights the suitability of HashEmAll for real-world ZKP applications involving large-scale data authentication. We also highlight the ability to translate the HashEmAll methodology to various ZK-friendly hash functions and different field sizes.
This paper comprehensively discusses the security mechanism of blockchain-based digital currency transactions, from the application of distributed ledgers, consensus algorithms, smart contracts, to the implementation of multiple signatures and cold storage schemes, to advanced privacy protection technologies, such as zero-knowledge proofs and homomorphic encryption. In particular, we delve into innovative mechanisms for attack prevention, including the fusion of heterogeneous multi-chain architectures with PoW+PoS hybrid consensus models. The article also details the practice of performance evaluation and security testing through a series of carefully designed experiments such as throughput and latency testing under different loads, resource utilization monitoring, and security reviews and comparisons with competitors. Experimental results show that the system exhibits good throughput growth under high load, but with the increase of latency, resource utilization is efficient and tends to saturation, most of the security protection mechanisms meet the standards, but access control problems that need to be optimized and high-risk vulnerabilities to be repaired are also found.
The rapid development of blockchain has significantly promoted research on zero-knowledge proofs (ZKPs), especially zero-knowledge succinct noninteractive arguments of knowledge (zk-SNARK). As is well known, protocol proof and verification time, as well as proof size, are the main obstacles that restrict the implementation of ZKPs in practical applications, so they have become the main concerns of researchers in recent years. This work achieves a new recursive zk-SNARK called GENES, which does not have a trusted setup and is secure under the standard discrete logarithm assumption. GENES is designed from the form of the rank-1 constraint system (R1CS) satisfiability problem. Recursive proof composition is achieved by merging multiple R1CS instances, which transforms the verification of numerous proofs into the verification of a single proof. Moreover, multi-helpers amortize proof commitments in this study, significantly reducing the computational pressure and time cost of proof generation. Compared with previous work, GENES effectively improves the proof time and verification time, but at the cost of larger proof sizes. We provide a blockchain Layer-1 scaling solution leveraging GENES to demonstrate its practicality.
Open access
Blockchain Technology Applications and Security
Advanced Steganography and Watermarking Techniques
While maternal mortality decreased during the Millennium Development Goals era, it remains unacceptably high, with stagnation in reductions possible due to shocks such as COVID-19. Most women in low- and middle-income countries already receive antenatal care and over half give birth in health facilities. In cities, use of health facilities for childbirth is near universal (>90%). Cities present complex challenges in ensuring pregnant women receive equitable, high-quality care. The UrbanBirth Collective is a portfolio of projects in sub-Saharan African cities seeking to address an important knowledge gap: how to adapt urban healthcare systems and lived environments to improve maternal and perinatal well-being? Its key focus is care during labour, childbirth, and the early postnatal period, when most poor maternal and perinatal outcomes occur. Our starting projects focus on harnessing open source data to examine and compare cities on the continent, including in-depth case studies of three cities: Grand Conakry (Guinea), Grand Nokoué metropolitan area (Benin), and Lubumbashi (Democratic Republic of the Congo), where we will capture and analyse three main dimensions of the dynamics: maternal health service provision; maternal healthcare use by women; and the complex, nonlinear interactions between the provision and use of care within the spatial, social, and political ecosystem of a city. By comparing these three cities, we shall propose a generalisable model which can be validated and applied in other cities in sub-Saharan Africa. The growth of cities demands increasing attention on future-proofing them with the capacity to develop, implement, and continuously adapt a coherent strategy for the provision of equitable maternal and newborn care. Our ambition is to contribute to reaching zero preventable maternal deaths in cities. To achieve these goals through understanding specific contexts and facilitating the adoption and application of research findings and recommendations, we will collaborate closely with local stakeholders, including healthcare workers, community leaders, and policymakers.
Abstract Authentication and access control for Cyber-Physical Systems (CPSs) are pivotal for protecting systems and their users from problems related to harmful actions and the malicious use of retrieved data. In some situations, making access decisions requires using user information, thereby challenging their privacy. Attribute-based access control (ABAC) supports dynamic and context-aware access decisions that are attractive in cyber-physical system environments. However, privacy preservation for access decisions is an open issue for authorization and is not supported by existing ABAC models. For example, if access decisions need to be made based on private attribute values such as health data, the corresponding access control policies need to be revealed. This paper reviews the ABAC, homomorphic encryption (HE), and zero-knowledge proof (ZKP) approaches, confirming the gap in privacy preservation in ABAC. Based on this observation, we further present the application of a new ZKP-based protocol in which ABAC allows for the privacy-preserving evaluation of attributes. This protocol is implemented and evaluated in terms of its performance and security. The evaluation demonstrates that there is a possibility for privacy-preserving ABAC, which may benefit the use of CPS, e.g., in underground and open-pit mines.
Secure multi-party computation is an area in cryptography which studies how multiple parties can compare their private information without revealing it. Besides digital protocols, many unconventional protocols for secure multi-party computation using physical objects have also been developed. The vast majority of them use playing cards as the main tools. In 2024, Kaneko et al. introduced the use of a balance scale and coins in zero-knowledge proof protocols for pencil puzzles. In this paper, we extend the use of these tools to secure multi-party computation. In particular, we develop four protocols that can securely compute any $n$-variable Boolean function using a balance scale and coins.
Low-Rank Adaptation (LoRA) is a widely adopted method for customizing large-scale language models. In distributed, untrusted training environments, an open source base model user may want to use LoRA weights created by an external contributor, leading to two requirements: (1) the base model user must confirm that the LoRA weights are effective when paired with the intended base model, and (2) the LoRA contributor must keep their proprietary weights private until compensation is assured. We present ZKLoRA, a zero-knowledge verification protocol that relies on succinct proofs and our novel Multi-Party Inference procedure to verify LoRA-base model compatibility without exposing LoRA weights. ZKLoRA produces deterministic correctness guarantees and validates each LoRA module in only 1-2 seconds on state-of-the-art large language models. This low-latency approach enables nearly real-time verification and promotes secure collaboration among geographically decentralized teams and contract-based training pipelines. The protocol ensures that the delivered LoRA module works as claimed, safeguarding the contributor's intellectual property while providing the base model user with verification of compatibility and lineage.
Open access
2 source records
Geophysical Methods and Applications
Robotics and Automated Systems
Target Tracking and Data Fusion in Sensor Networks
Recently, the practical applications of advanced cryptographic protocols, such as Multi-Party Computation (MPC), Fully Homomorphic Encryption (FHE), and Zero Knowledge Proofs (ZKP), have spurred the development of a series of new symmetric encryption primitives. These novel symmetric encryption primitives, referred to as Arithmetization-Oriented (AO) ciphers, aim to minimize the number of field multiplications in large finite fields, including 𝔽2n or 𝔽p. In order to evaluate the algebraic degrees of AO ciphers over 𝔽2n, the general monomial prediction technique was proposed by Cui et al. at ASIACRYPT 2022. However, when using their searching tool to evaluate the algebraic degrees of AO ciphers with complex affine layers, the efficiency is low, preventing solutions within a predetermined timeframe. In this study, we extend the propagation rules of monomials for field-based operations and present an automatic searching tool based on Mixed Integer Linear Programming (MILP) and Boolean Satisfiability (SAT) Problem for evaluating the upper bound of the algebraic degrees. Moreover, to accurately calculate the algebraic degrees of monomials in the SAT model, we improve the sequence encoding method, enabling it to accurately determine whether the monomials of degree d exist in the output. We apply our new searching tool to various AO ciphers, including Chaghri, MiMC, and Ciminion. For Chaghri, we compare our results with the Coefficient Grouping technique proposed by Liu et al. at EUROCRYPT 2023, and our results yield tighter upper bounds compared to Liu et al.’s findings. Additionally, we evaluate the algebraic degrees of Chaghri and MiMC with arbitrary complex affine layers and obtain tighter bounds compared to the results from Liu et al. at CRYPTO 2023. Regarding Ciminion, we have observed that starting from the 4-th round, the upper bound on the algebraic degrees for each round actually 1 degree lower than the previous bound. Our searching tool enables a more precise evaluation of the algebraic degrees of AO ciphers, contributing to a deeper understanding of the design and analysis of such primitives.
Solving Alpha — Version 5.2 The derivation of the fine structure constant, closed from two further directions. The fine structure constant α was derived from the self-reference axiom σ = 1/(1+σ) in the first paper of this series and reaffirmed across Versions 1 through 4. The Pentagon formula α⁻¹ = 360/φ² − 2/φ³ + 1/(3⁵φ⁵) + 1/(7⁷φ⁷) reproduces the Morel 2020 atomic recoil determination of α⁻¹ = 137.035999206(11) to within 0.05σ, with zero free parameters and no experimental input. That derivation stands as originally posted. Version 5.2 does not derive α again. It closes the proof from two further directions, each structurally independent of the original derivation and of each other. The first closure is internal uniqueness. Within a pre-specified coefficient pool drawn from the irreducible representations of the binary icosahedral group, the spectral structure of the 600-cell polytope, and the self-referential reciprocal-power family — defined before the formula is consulted and requiring no knowledge of α — the Pentagon formula is the unique 1σ match to Morel 2020. The nearest structurally distinct competitor sits 139× further from the measured value. The four prime exponents (2, 3, 5, 7) of the formula are independently attested by the seventh spectral moment of the 600-cell adjacency matrix, μ₇ = Tr(A⁷)/1440 = 50,400 = 2⁵ · 3² · 5² · 7. The Pentagon formula is not one of many φ-series that fit; it is the only structurally admissible one. The second closure is external overdetermination. The same number α⁻¹ = 137.036 that the Pentagon formula produces is independently recovered, with no electromagnetic input, from three disconnected non-electromagnetic sectors. The cosmological constant Λ from Planck CMB and BAO, the gravitational coupling G from CODATA torsion balance measurements, and the Hubble expansion rate H₀ from SH0ES distance ladders all sit on a single straight line whose slope is α⁻¹ and whose intercept is φ⁻². The horizontal coordinates of that line are forced by Dirichlet's 1837 class number theorem for the field ℚ(√5). Four disconnected experimental programmes, four independent determinations of α⁻¹, one common value. The original derivation gave the number. The first closure shows that no other formula in the structurally admissible space gives that number. The second closure shows that the same number is the unique slope on which four disconnected experimental sectors agree. The proof was complete in V1; it is now closed on three sides. The asymptotic series for α⁻¹ is presented in fully derived form, with coefficient C_k = 2^(k²) counting the directed coupling configurations among k self-referential modes at maximum entropy equilibrium. The series shares the asymptotic character of QED's own perturbation expansion, with optimal truncation near k = 6 settling within 1.65σ of the most precise measurement. A fifth term is pre-registered before any measurement at the required precision exists to test it. Confirmation of either the Parker 2018 caesium or Fan 2023 electron g−2 determinations as the correct value of α⁻¹ at high significance falsifies the formula at the current truncation order; the framework commits to Morel 2020 as the correct value. The fine structure constant is a theorem of self-referential geometry on the field ℚ(√5). The original derivation, the internal uniqueness closure, and the external overdetermination closure are now on the public record together. Ten revisions between V5 and V5.2 are documented inline; the bone-structure claims survive intact. Supplementary ablation scripts and machine-readable results are deposited alongside this record for full reproducibility. Keywords: fine structure constant, self-reference, 600-cell, binary icosahedral group, Dirichlet class number, asymptotic series, Pentagon Physics, derivation closure, falsifiable prediction, ℚ(√5)
As an emerging paradigm in digital identity, Decentralized Identity (DID) appears advantages over traditional identity management methods in a variety of aspects, e.g., enhancing user-centric online services and ensuring complete user autonomy and control. Verifiable Credential (VC) techniques are used to facilitate decentralized DID-based access control across multiple entities. However, existing DID schemes generally rely on a distributed public key infrastructure that also causes challenges, such as context information deduction, key exposure, and issuer data leakage. To address the issues above, this paper proposes a issuer-hiding and privacy-preserving DID multi-party authentication model with a signature-less VC scheme, named SLVC-DIDA, for the first time. Our proposed scheme avoids the dependence on signing keys by employing hashing and issuer membership proofs, which supports universal zero-knowledge multi-party DID authentications, eliminating additional technical integrations. We adopt a novel zero-knowledge circuit to maintain the anonymity of the issuer set, thereby enabling public verification while safeguarding the privacy of identity attributes via a Merkle tree-based VC list. Furthermore, by eliminating reliance on a Public Key Infrastructure (PKI), SLVC-DIDA enables decentralized and self-sovereign DID authentication. Our experiments further evaluate the effectiveness and practicality of SLVC-DIDA.
We study formally determined inverse problems with passive measurements for one dimensional evolution equations where the goal is to simultaneously determine both the initial data as well as the variable coefficients in such an equation from the measurement of its solution at a fixed spatial point for a certain amount of time. This can be considered as a one-dimensional model of widely open inverse problems in photo-acoustic and thermo-acoustic tomography. We provide global uniqueness results for wave and heat equations stated on bounded or unbounded spatial intervals. Contrary to all previous related results on the subject, we do not impose any genericity assumptions on the coefficients or initial data. Our proofs are based on creating suitable links to the well understood spectral theory for 1D Schrödinger operators. In particular, in the more challenging case of a bounded spatial domain, our proof for the inverse problem partly relies on the following two ingredients, namely (i) a Paley-Wiener type theorem for Schrödinger operators due to Remling \cite{Remling2002SchrdingerOA} and a theorem of Levinson \cite{Levinson1940} on distribution of zeros of entire functions of regular growth that together provide a quantifiable link between support of a compactly supported function and the upper density of its vanishing Schrödinger spectral modes and (ii) a result of Gesztesy and Simon \cite{Gesztesy1999InverseSA} on partial data inverse spectral problems for reconstructing an unknown potential in a 1D Schrödinger operator from the knowledge of only a fraction of its spectrum.
Yang Liu, Ru Huo, Ningjie Gao, Cheng Chi · 5 authors
In order to address the challenges encountered in the current Industrial Internet of Things scenarios, such as single points of failure, difficulties in ensuring data privacy and integrity, and a lack of access control, a blockchain-based data security exchange architecture was proposed. To ensure the privacy of industrial data, a data exchange process based on public key encryption and keyword search was introduced. Industrial data is encrypted multiple times and uploaded to the blockchain network. Users retrieve ciphertext from the cloud server after obtaining the key through the blockchain and then decrypt it. To achieve flexible access control, a zero-knowledge proof-based access control mechanism was proposed, utilizing Pedersen commitments and zero-knowledge proofs for access permission issuance, validation, and revocation. Additionally, various forms of smart contracts were proposed for secure data exchange, user authentication, access authorization, and data integrity verification. Finally, a system prototype was built and experimental results confirmed the superiority of the proposed approach.
Currently, PAKE (Password Authenticated Key Exchange) protocols on lattice using a single-server architecture are widely applied. However, such protocols are vulnerable to server leakage attacks, dictionary attacks, and other threats. To address these issues, researchers have proposed multi-server and two-server architecture-based PAKE protocols. However, PAKE protocols in a multi-server architecture require the use of complex cryptographic primitives such as signatures, and zero-knowledge proofs to ensure security, which reduces the execution efficiency of the protocol. To solve these problems, we propose a two-server PAKE protocol on the lattice based on the MLWE (Module learning with errors) problem. The protocol is built using instances of the MLWE problem and utilizes the Peikert error coordination technique, which ensures both parties with similar values arrive at the same result through computation. Additionally, we introduce the error pairing hypothesis and demonstrates its security within the random oracle model. The protocol securely stores different shares of password information across various servers. This approach protects user password data, even if one of the servers is compromised. Compared to similar protocols, we avoid the use of numerous cryptographic primitives, and can better resist quantum computing attacks and server leakage. And we reduce computational and communication costs, and can better meet practical application needs.
The emergence of quantum computing has provided new paradigms for cryptography. On the one hand, it poses significant new threats to existing classically cryptographic systems, requiring the community to define new security models that capture what a quantum adversary can do. On the other hand, it gives us new tools to design cryptographic protocols, with weaker assumptions than in the classical world, or even protocols that are impossible classically. In this survey, we first give an overview of new security definitions for classical cryptography, considering quantum adversaries who can either only use local quantum computation (post-quantum security), or even send quantum messages and in particular have access to oracle in superposition (quantum security). We explore these new notions through the examples of commitments, zero-knowledge proofs, encryption, and signatures. Then, we present what is arguably the most famous application of quantum cryptography: quantum key distribution (QKD) protocols that take advantage of unique properties of quantum mechanics to provide secure communication unconditionally. We also explore cryptography beyond QKD, focusing on unclonable cryptography: a family of cryptographic functionalities, built with quantum states, and designed to be resistant to counterfeit by leveraging the “no-cloning” theorem. We examine in particular quantum money, but also the recent notions of unclonable encryption and copy-protection, including related variants. By presenting a comprehensive survey of these topics, this paper aims to provide a thorough understanding of the current landscape and future potential of quantum cryptography.
Ilia Shumailov, Daniel Ramage, Sarah Meiklejohn, Peter Kairouz · 7 authors
We often interact with untrusted parties. Prioritization of privacy can limit the effectiveness of these interactions, as achieving certain goals necessitates sharing private data. Traditionally, addressing this challenge has involved either seeking trusted intermediaries or constructing cryptographic protocols that restrict how much data is revealed, such as multi-party computations or zero-knowledge proofs. While significant advances have been made in scaling cryptographic approaches, they remain limited in terms of the size and complexity of applications they can be used for. In this paper, we argue that capable machine learning models can fulfill the role of a trusted third party, thus enabling secure computations for applications that were previously infeasible. In particular, we describe Trusted Capable Model Environments (TCMEs) as an alternative approach for scaling secure computation, where capable machine learning model(s) interact under input/output constraints, with explicit information flow control and explicit statelessness. This approach aims to achieve a balance between privacy and computational efficiency, enabling private inference where classical cryptographic solutions are currently infeasible. We describe a number of use cases that are enabled by TCME, and show that even some simple classic cryptographic problems can already be solved with TCME. Finally, we outline current limitations and discuss the path forward in implementing them.
Abstract— Remote electronic voting promises increased accessibility but remains constrained by persistent challenges related to coercion in unsupervised environments, credential compromise, and the difficulty of sustaining long-term voter trust. While coercion-resistant approaches commonly rely on revoting, most existing systems treat credential loss or recovery as an administrative exception, often reintroducing identity linkage or trusted intermediaries and offering limited means for voters or observers to verify that an election unfolded as intended. This paper presents Arcaunt, a remote voting architecture that elevates anonymous credential recovery to a first-class security property and integrates it directly into the voting lifecycle. The architecture introduces an Anonymous Recovery Channel (ARC), enabling voters to revoke and replace compromised credentials without identity disclosure or reliance on administrator discretion. Recovery is logically and operationally decoupled from ballot casting. This preserves ballot secrecy and prevents temporary compromise of credentials, devices, or voter autonomy from becoming a permanent loss of voting control. Arcaunt builds on established cryptographic mechanisms, including publicly verifiable bulletin boards, commitment-based ballots, and unlinkable bearer credentials. These components provide ballot integrity and verifiability without exposing voter identities and form the foundation on which revoting, recovery, and auditability are composed. Individual assurance is provided through deferred, non-transferable verification mechanisms: voters receive a receipt at ballot submission, while verification becomes possible only after election closure, preventing real-time feedback that could enable coercion while still allowing voters to confirm that their final valid ballot was recorded and included. At the system level, integrity is enforced through an append-only, publicly auditable ledger and deterministic “last valid vote” counting rules, ensuring that administrative database access cannot alter election outcomes without detection. The architecture explicitly bounds its threat model, acknowledging limits against global traffic analysis and continuous coercion while constraining failures to be temporary and non-scalable. We analyze the security properties of the proposed system under realistic adversarial assumptions and evaluate a prototype implementation, demonstrating that anonymous recovery, coercion-resistant revoting, individual verification, and public auditability can be combined efficiently without reliance on trusted administrators or specialized hardware. Keywords—e-voting, arcaunt, anonymous recovery channel (arc), coercion resistance, sha-3, digital democracy, govtech, zero-knowledge proofs.
Fully homomorphic encryption (FHE) and zero-knowledge proofs (ZKPs) are emerging as solutions for data security in distributed environments. However, the widespread adoption of these encryption techniques is hindered by their significant computational overhead, primarily resulting from core cryptographic operations that involve large integer arithmetic. This paper presents a formalization of multi-word modular arithmetic (MoMA), which breaks down large bit-width integer arithmetic into operations on machine words. We further develop a rewrite system that implements MoMA through recursive rewriting of data types, designed for compatibility with compiler infrastructures and code generators. We evaluate MoMA by generating cryptographic kernels, including basic linear algebra subprogram (BLAS) operations and the number theoretic transform (NTT), targeting various GPUs. Our MoMA-based BLAS operations outperform state-of-the-art multi-precision libraries by orders of magnitude, and MoMA-based NTTs achieve near-ASIC performance on commodity GPUs.
Bit commitment is a fundamental cryptographic primitive and a cornerstone for numerous two-party cryptographic protocols, including zero-knowledge proofs. However, it has been proven that unconditionally secure bit commitment, both classical and quantum, is impossible. In this work, we demonstrate that imposing a restriction on the committing party to perform only separable operations enables secure quantum bit commitment schemes. Specifically, we prove that in any perfectly hiding bit commitment protocol, an honestly-committing party limited to separable operations will be detected with high probability if they attempt to alter their commitment. To illustrate our findings, we present an example protocol.
Non-interactive zero-knowledge (NIZK) proof systems are a cornerstone of modern cryptography, but their security has received little attention in the quantum settings. Motivated by improving our understanding of this fundamental primitive against quantum adversaries, we propose a new definition of security against quantum adversary. Specifically, we define the notion of quantum simulation soundness (SS-NIZK), that allows the adversary to access the simulator in superposition. We show a separation between post-quantum and quantum security of SS-NIZK, and prove that Sahai’s construction for SS-NIZK (in the CRS model) can be made quantumly-simulation-sound. As an immediate application of our new notion, we prove the security of the Naor-Yung paradigm in the quantum settings, with respect to a strong quantum IND-CCA security notion. This provides the quantum analogue of the classical dual key approach to prove the security of encryption schemes. Along the way, we introduce a new notion of quantum-query advantage functions, which may be used as a general framework to show classical/quantum separation for other cryptographic primitives, and it may be of independent interest.
The convergence of artificial intelligence and blockchain technology offers a compelling paradigm for deploying secure, auditable, and decentralisedintelligent systems in distributed environments where trust between participants cannot be assumed. Existing AI deployment frameworks lackimmutable audit trails, transparent model governance, and cryptographic integrity guarantees — requirements increasingly mandated by regulatoryframeworks including the EU AI Act and GDPR. This study presents ChainMind, a blockchain-enabled secure intelligent system frameworkintegrating smart contract-governed model lifecycle management, federated learning with on-chain gradient verification, and zero-knowledge proof(ZKP)-based inference auditing for privacy-preserving accountability. ChainMind was deployed and evaluated across three distributed intelligentsystem applications: a decentralised medical AI consortium (6 European hospitals, 284,000 patient records), a cross-border financial fraud detectionnetwork (4 banks, Germany and Italy), and a smart city data marketplace (Stuttgart urban IoT network, 12,400 sensors). ChainMind achieved modeltampering detection accuracy of 99.97%, federated learning convergence within 18.3% fewer rounds than standard FedAvg under adversarialgradient poisoning, and ZKP inference verification latency of 47.3 ms — compatible with real-time deployment. These results establish ChainMind asa technically viable and regulatory-compliant framework for blockchain-enabled secure AI in distributed environments.
Folding schemes are an exciting new primitive, transforming the task of performing multiple zero-knowledge proofs of knowledge for a relation into performing just one zero-knowledge proof, for the same relation, and a number of cheap inclusion-proofs. Recently, folding schemes have been used to amortize the cost associated with proving different statements to multiple distinct verifiers, which has various applications. We observe that for these uses, leaking information about the statements folded together can be problematic, yet this happens with previous constructions. Towards resolving this issue, we give a natural definition of privacy preserving folding schemes, and what security they should offer. To construct privacy preserving folding schemes, we first define statement hiders, a primitive which might be of independent interest. In a nutshell, a statement hider hides an instance of a relation as a new instance in the same relation. The new instance is in the relation if and only if the initial instance is. With this building block, we can utilize existing folding schemes to construct a privacy preserving folding scheme, by first hiding each of the statements. Folding schemes allow verifying that a statement was folded into another statement, while statement hiders allow verifying that a statement was hidden as another statement.
With the rapid development of quantum computers, proofs of quantumness have recently become an interesting and intriguing research direction. However, in all current schemes for proofs of quantumness, quantum provers almost invariably face the risk of being maliciously exploited by classical verifiers. In fact, through malicious strategies in interaction with quantum provers, classical verifiers could solve some instances of hard problems that arise from the specific scheme in use. In other words, malicious verifiers can break some schemes (that quantum provers are not aware of) through interaction with quantum provers. All this is due to the lack of formalization that prevents malicious verifiers from extracting useful information in proofs of quantumness. To address this issue, we formalize zero-knowledge proofs of quantumness. Intuitively, the zero-knowledge property necessitates that the information gained by the classical verifier from interactions with the quantum prover should not surpass what can be simulated using a simulated classical prover interacting with the same verifier. As a result, the new zero-knowledge notion can prevent any malicious verifier from exploiting quantum advantage. Interestingly, we find that the classical zero-knowledge proof is sufficient to compile some existing proofs of quantumness schemes into zero-knowledge proofs of quantumness schemes. Due to some technical reason, it appears to be more general to require zero-knowledge proof on the verifier side instead of the prover side. Intuitively, this helps to regulate the verifier's behavior from malicious to be honest-but-curious. As a result, both parties will play not only one role in the proofs of quantumness but also the dual role in the classical zero-knowledge proof. Specifically, the two principle proofs of quantumness schemes: Shor's factoring-based scheme and learning with errors-based scheme in [Brakerski et al, FOCS, 2018], can be transformed into zero-knowledge proofs of quantumness by requiring an extractable non-interactive zero-knowledge argument on the verifier side. Notably, the zero-knowledge proofs of quantumness can be viewed as an enhanced security notion for proofs of quantumness. To prevent malicious verifiers from exploiting the quantum device's capabilities or knowledge, it is advisable to transition existing proofs of quantumness schemes to this framework whenever feasible.
The advancement of AI models, especially those powered by deep learning, faces significant challenges in data-sensitive industries like healthcare and finance due to the distributed and private nature of data. Federated Learning (FL) and Secure Federated Learning (SFL) enable collaborative model training without data sharing, enhancing privacy by encrypting shared intermediate results. However, SFL currently lacks effective Byzantine robustness, a critical property that ensures model performance remains intact even when some participants act maliciously. Existing Byzantine-robust methods in FL are incompatible with SFL due to the inefficiency and limitations of encryption operations in handling complex aggregation calculations. This creates a significant gap in secure and robust model training. To address this gap, we propose ByzSFL, a novel SFL system that achieves Byzantine-robust secure aggregation with high efficiency. Our approach offloads aggregation weight calculations to individual parties and introduces a practical zero-knowledge proof (ZKP) protocol toolkit. This toolkit supports widely used operators for calculating aggregation weights, ensuring correct computations without compromising data privacy. Not only does this method maintain aggregation integrity, but it also significantly boosts computational efficiency, making ByzSFL approximately 100 times faster than existing solutions. Furthermore, our method aligns with open-source AI trends, enabling plaintext publication of the final model without additional information leakage, thereby enhancing the practicality and robustness of SFL in real-world applications.