International Association
for Cryptologic Research

We show that the smallness of message spaces can be used as a checksum allowing to hedge against CCA1 attacks in additively homomorphic encryption schemes. We first show that the additively homomorphic variant of Damgård's Elgamal provides IND-CCA1 security under the standard DDH assumption. Earlier proofs either required non-standard assumptions or only applied to hybrid versions of Damgård's Elgamal, which are not additively homomorphic. Our security proof builds on hash proof systems and exploits the fact that encrypted messages must be contained in a polynomial-size interval in order to enable decryption. With $3$ group elements per ciphertext, this positions Damgård's Elgamal as the most efficient/compact DDH-based additively homomorphic CCA1 cryptosystem. Under the same assumption, the best candidate so far was the lite Cramer-Shoup cryptosystem, where ciphertexts consist of $4$ group elements. We extend this observation to build an IND-CCA1 variant of the Boneh-Goh-Nissim encryption scheme, which allows evaluating 2-DNF formulas on encrypted data. By computing tensor products of Damgård's Elgamal ciphertexts, we obtain product ciphertexts consisting of $9$ group elements (instead of $16$ elements if we were tensoring lite Cramer-Shoup ciphertexts) in the target group of a bilinear map. Using similar ideas, we also obtain a CCA1 variant of the Elgamal-Paillier cryptosystem by forcing $\lambda$ plaintext bits to be zeroes, which yields CCA1 security almost for free. In particular, the message space remains exponentially large and ciphertexts are as short as in the IND-CPA scheme. We finally adapt the technique to the Castagnos-Laguillaumie system.

In this work, we propose a construction for $ Multi~Input~Inner ~Product ~Encryption$ (MIPFE) that can handle vectors of variable length in different encryption slots. This construction is the first of its kind, as all existing MIPFE schemes allow only equal length vectors. The scheme is constructed in the private key setting, providing privacy for both message as well as the function, thereby achieving the so-called $full-hiding$ security. Our MIPFE scheme uses bilinear groups of prime order and achieves security under well studied cryptographic assumptions, namely, the symmetric external Diffie-Hellman assumption.

We analyse the cryptographic protocols underlying Delta Chat, a decentralised messaging application which uses e-mail infrastructure for message delivery. It provides end-to-end encryption by implementing the Autocrypt standard and the SecureJoin protocols, both making use of the OpenPGP standard. Delta Chat's adoption by categories of high-risk users such as journalists and activists, but also more generally users in regions affected by Internet censorship, makes it a target for powerful adversaries. Yet, the security of its protocols has not been studied to date. We describe five new attacks on Delta Chat in its own threat model, exploiting cross-protocol interactions between its implementation of SecureJoin and Autocrypt, as well as bugs in rPGP, its OpenPGP library. The findings have been disclosed to the Delta Chat team, who implemented fixes.

In a non-zero inner product encryption (NIPE) scheme, ciphertexts and keys are associated with vectors from some inner-product space. Decryption of a ciphertext for $\vec{x}$ is allowed by a key for $\vec{y}$ if and only if the inner product $\langle{\vec{x}},{\vec{y}}\rangle \neq 0$. Existing constructions of NIPE assume the length of the vectors are fixed apriori. We present the first constructions of $ unbounded $ non-zero inner product encryption (UNIPE) with constant sized keys. Unbounded here refers to the size of vectors not being pre-fixed during setup. Both constructions, based on bilinear maps, are proven selectively secure under the decisional bilinear Diffie-Hellman (DBDH) assumption.

Our constructions are obtained by transforming the unbounded inner product functional encryption (IPFE) schemes of Dufour-Sans and Pointcheval (ACNS 2019), one in the $strict ~ domain$ setting and the other in the $permissive ~ domain$ setting. Interestingly, in the latter case, we prove security from DBDH, a static assumption while the original IPE scheme relied on an interactive parameterised assumption. In terms of efficiency, features of the IPE constructions are retrained after transformation to NIPE. Notably, the public key and decryption keys have constant size.

Shortening the argument (three group elements or 1536 / 3072 bits over the BLS12-381/BLS24-509 curves) of the Groth16 zk-SNARK for R1CS is a long-standing open problem. We propose a zk-SNARK Polymath for the Square Arithmetic Programming constraint system using the KZG polynomial commitment scheme. Polymath has a shorter argument (1408 / 1792 bits over the same curves) than Groth16. At 192-bit security, Polymath's argument is nearly half the size, making it highly competitive for high-security future applications. Notably, we handle public inputs in a simple way. We optimized Polymath's prover through an exhaustive parameter search. Polymath's prover does not output $\mathbb{G}_{2}$ elements, aiding in batch verification, SNARK aggregation, and recursion. Polymath's properties make it highly suitable to be the final SNARK in SNARK compositions.

For more than forty years, two principal questions have been asked when designing verifiable election systems: how will the integrity of the results be demonstrated and how will the privacy of votes be preserved? Many approaches have been taken towards answering the first question such as use of MixNets and homomorphic tallying. But in the academic literature, the second question has always been answered in the same way: decryption capabilities are divided amongst multiple independent “trustees” so that a collusion is required to compromise privacy.

In practice, however, this approach can be fairly challenging to deploy. Human trustees rarely have a clear understanding of their responsibilities, and they typically all use identical software for their tasks. Rather than exercising independent judgment to maintain privacy, trustees are often reduced to automata who just push the buttons they are told to when they are told to, doing little towards protecting voter privacy. This paper looks at several aspects of the trustee experience. It begins by discussing various cryptographic protocols that have been used for key generation in elections, explores their impact on the role of trustees, and notes that even the theory of proper use of trustees is more challenging than it might seem. This is illustrated by showing that one of the only references defining a full threshold distributed key generation (DKG) for elections defines an insecure protocol. Belenios claims to rely on that reference for its DKG and security proof. Fortunately, it does not inherit the same vulnerability. We offer a security proof for the Belenios DKG.

The paper then discusses various practical contexts, in terms of humans, software, and hardware, and their impact on the practical deployment of a trustee-based privacy model.

End-to-End (E2E) encrypted messaging, which prevents even the service provider from learning communication contents, is gaining popularity. Since users care about maintaining access to their data even if their devices are lost or broken or just replaced, these systems are often paired with cloud backup solutions: Typically, the user will encrypt their messages with a fixed key, and upload the ciphertexts to the server. Unfortunately, this naive solution has many drawbacks. First, it often undermines the fancy security guarantees of the core application, such as forward secrecy (FS) and post-compromise security (PCS), in case the single backup key is compromised. Second, they are wasteful for backing up conversations in large groups, where many users are interested in backing up the same sequence of messages.

Instead, we formalize a new primitive called Compact Key Storage (CKS) as the "right" solution to this problem. Such CKS scheme allows a mutable set of parties to delegate to a server storage of an increasing set of keys, while each client maintains only a small state. Clients update their state as they learn new keys (maintaining PCS), or whenever they want to forget keys (achieving FS), often without the need to interact with the server. Moreover, access to the keys (or some subset of them) can be efficiently delegated to new group members, who all efficiently share the same server's storage.

We carefully define syntax, correctness, privacy, and integrity of CKS schemes, and build two efficient schemes provably satisfying these notions. Our line scheme covers the most basic "all-or-nothing" flavor of CKS, where one wishes to compactly store and delegate the entire history of past secrets. Thus, new users enjoy the efficiency and compactness properties of the CKS only after being granted access to the entire history of keys. In contrast, our interval scheme is only slightly less efficient but allows for finer-grained access, delegation, and deletion of past keys.

There has been significant progress over the past seven years in model reverse engineering (RE) for neural network (NN) hardware. Although there has been systematization of knowledge (SoK) in an overall sense, however, the treatment from the hardware perspective has been far from adequate. To bridge this gap, this paper systematically categorizes the types of NN hardware used prevalently by the industry/academia, and also the model RE attacks/defenses published in each category. Further, we sub-categorize existing NN model RE attacks based on different criteria including the degree of hardware parallelism, threat vectors like side channels, fault-injection, scan-chain attacks, system-level attacks, type of asset under attack, the type of NN, exact versus approximate recovery, etc.

We make important technical observations and identify key open research directions. Subsequently, we discuss the state-of-the-art defenses against NN model RE, identify certain categorization criteria, and compare the existing works based on these criteria. We note significant qualitative gaps for defenses, and suggest recommendations for important open research directions for protection of NN models. Finally, we discuss limitations of existing work in terms of the types of models where security evaluation or defenses were proposed, and suggest open problems in terms of protecting practically expensive model IPs.

In the useful and well studied model of secret-sharing schemes, there are $n$ parties and a dealer, which holds a secret. The dealer applies some randomized algorithm to the secret, resulting in $n$ strings, called shares; it gives the $i$'th share to the $i$'th party. There are two requirements. (1) correctness: some predefined subsets of the parties can jointly reconstruct the secret from their shares, and (2) security: any other set gets no information on the secret. The collection of predefined qualified sets is called an access structure (AS).

This model assumes that the number of parties is known when preparing the shares and giving the shares to the parties; furthermore, the sharing algorithm and the share size are determined by the number of parties, e.g. in the best-known secret-sharing scheme for an arbitrary $n$-party access structure the share size is $1.5^{n}$ by Appelbaum and Nir.

The assumption that the number of parties is known in advance is problematic in many scenarios. Of course, one can take some upper bound on the number of parties. On one hand, if this bound is big, then the share size will be large even if only few parties actually participate in the scheme. On the other hand, if this bound is small, then there is a risk that too many parties will arrive and no further shares can be produced; this will require an expensive re-sharing of the secret and updating all shares (which can be impossible if some parties are temporally off-line). Thus, we need to consider models with an unbounded number of parties.

To address these concrens, Komargodski, Naor, and Yogev defined \emph{evolving secret-sharing schemes} with an unbounded number of parties. In a nutshell, evolving AS's are defined as a monotone collection of finite qualified sets, such that at any time $t$ a set $A\subseteq [t]$ is either qualified or not, depending only on $A$ itself, and not on $t$ (a `global' monotonicity notion).

Quantum secret sharing (QSS) in the standard $n$-party setting, where the secret is an arbitrary quantum state (say, qbit), rather than classical data. In face of recent advancements in quantum computing, this is a natural notion to consider, and has been studied before.

In this work, we explore the natural notion of quantum evolving secret sharing (QESS). While this notion has been studied by Samadder 20', we make several new contributions. (1) The notion of QESS was only implicit in the above work. We formalize this notion (as well as AS's for which it is applicable), and in particular argue that the variant implied by the above work did not require `global monotonicity' of the AS, which was the standard in the evolving secret sharing literature, and appears to be useful for QESS as well. (2) Discuss the applicability and limitations of the notion in the quantum setting that follow from the no-cloning theorem, and make its usability more limited. Yet, we argue that fundamental advantages of the evovling setting, such as keeping parties' shares independent of the total number of parties that arrive can be mantainted in the quantum setting. (3) We characterize the AS's ammenable to construction of QSSS - so called `no cloning' evolving AS's, and point out that this class is not severly restricted relatively to the class of all evolving AS's. On the positive side, our construction combines the compiler of [Smith 00'] with ideas of hybrid secret sharing of [Goyal et. al 23'], to obtain a construction with share size comparable to the best classical linear share complexity of the scheme.

Cryptographic hash functions are said to be the work-horses of modern cryptography. One of the strongest approaches to assess a cryptographic hash function's security is indifferentiability. Informally, indifferentiability measures to what degree the function resembles a random oracle when instantiated with an ideal underlying primitive. However, proving the indifferentiability security of hash functions has been challenging due to complex simulator designs and proof arguments. The Sponge construction is one of the prevalent hashing method used in various systems. The Sponge has been shown to be indifferentiable from a random oracle when initialized with a random permutation.

In this work, we first introduce $\mathsf{GSponge}$, a generalized form of the Sponge construction offering enhanced flexibility in input chaining, field sizes, and padding types. $\mathsf{GSponge}$ not only captures all existing sponge variants but also unveils new, efficient ones. The generic structure of $\mathsf{GSponge}$ facilitates the discovery of two micro-optimizations for already deployed sponges. Firstly, it allows a new padding rule based on zero-padding and domain-separated inputs, saving one full permutation call in certain cases without increasing the generation time of zero-knowledge proofs. Secondly, it allows to absorb up to $\mathsf{c}/2$ more elements (that can save another permutation call for certain message lengths) without compromising the indifferentiability security. These optimizations enhance hashing time for practical use cases such as Merkle-tree hashing and short message processing.

We then propose a new efficient instantiation of $\mathsf{GSponge}$ called $\mathsf{Sponge2}$ capturing these micro-optimizations and provide a formal indifferentiability proof to establish both $\mathsf{Sponge2}$ and $\mathsf{GSponge}$'s security. This proof, simpler than the original for Sponges, offers clarity and ease of understanding for real-world practitioners. Additionally, it is demonstrated that $\mathsf{GSponge}$ can be safely instantiated with permutations defined over large prime fields, a result not previously formally proven.

$\mathrm{SPHINCS^{+}}$ is a post-quantum signature scheme that, at the time of writing, is being standardized as $\mathrm{SLH\text{-}DSA}$. It is the most conservative option for post-quantum signatures, but the original tight proofs of security were flawed—as reported by Kudinov, Kiktenko and Fedorov in 2020. In this work, we formally prove a tight security bound for $\mathrm{SPHINCS^{+}}$ using the EasyCrypt proof assistant, establishing greater confidence in the general security of the scheme and that of the parameter sets considered for standardization. To this end, we reconstruct the tight security proof presented by Hülsing and Kudinov (in 2022) in a modular way. A small but important part of this effort involves a complex argument relating four different games at once, of a form not yet formalized in EasyCrypt (to the best of our knowledge). We describe our approach to overcoming this major challenge, and develop a general formal verification technique aimed at this type of reasoning. Enhancing the set of reusable EasyCrypt artifacts previously produced in the formal verification of stateful hash-based cryptographic constructions, we (1) improve and extend the existing libraries for hash functions and (2) develop new libraries for fundamental concepts related to hash-based cryptographic constructions, including Merkle trees. These enhancements, along with the formal verification technique we develop, further ease future formal verification endeavors in EasyCrypt, especially those concerning hash-based cryptographic constructions.

One of the main issues to deal with for fully homomorphic encryption is the noise growth when operating on ciphertexts. To some extent, this can be controlled thanks to a so-called gadget decomposition. A gadget decomposition typically relies on radix- or CRT-based representations to split elements as vectors of smaller chunks whose inner products with the corresponding gadget vector rebuilds (an approximation of) the original elements. Radix-based gadget decompositions present the advantage of also supporting the approximate setting: for most homomorphic operations, this has a minor impact on the noise propagation but leads to substantial savings in bandwidth, memory requirements and computational costs. A typical use-case is the blind rotation as used for example in the bootstrapping of the TFHE scheme. On the other hand, CRT-based representations are convenient when machine words are too small for directly accommodating the arithmetic on large operands. This arises in two typical cases: (i) in the hardware case with multipliers of restricted size, e.g., 17 bits; (ii) in the software case for ciphertext moduli above, e.g., 128 bits.

This paper presents new CRT-based gadget decompositions for the approximate setting, which combines the advantages of non-exact decompositions with those of CRT-based decompositions. Significantly, it enables certain hardware or software realizations otherwise hardly supported like the two aforementioned cases. In particular, we show that our new gadget decompositions provide implementations of the (programmable) bootstrapping in TFHE relying solely on native arithmetic and offering extra degrees of parallelism.

In this note, we present additional preliminary analysis dedicated to Ascon-Xof and Ascon-Hash [DEMS19].

This paper presents an optimization of the memory cost of the quantum Information Set Decoding (ISD) algorithm proposed by Bernstein (PQCrypto 2010), obtained by combining Prange's ISD with Grover's quantum search.

When the code has constant rate and length $n$, this algorithm essentially performs a quantum search which, at each iterate, solves a linear system of dimension $\mathcal{O}(n)$. The typical code lengths used in post-quantum public-key cryptosystems range from $10^3$ to $10^5$. Gaussian elimination, which was used in previous works, needs $\mathcal{O}(n^2)$ space to represent the matrix, resulting in millions or billions of (logical) qubits for these schemes.

In this paper, we propose instead to use the algorithm for sparse matrix inversion of Wiedemann (IEEE Trans. inf. theory 1986). The interest of Wiedemann's method is that one relies only on the implementation of a matrix-vector product, where the matrix can be represented in an implicit way. This is the case here.

We propose two main trade-offs, which we have fully implemented, tested on small instances, and benchmarked for larger instances. The first one is a quantum circuit using $\mathcal{O}(n)$ qubits, $\mathcal{O}(n^3)$ Toffoli gates like Gaussian elimination, and depth $\mathcal{O}(n^2 \log n)$. The second one is a quantum circuit using $\mathcal{O}(n \log^2 n)$ qubits, $\mathcal{O}(n^3)$ gates in total but only $\mathcal{O}( n^2 \log^2 n)$ Toffoli gates, which relies on a different representation of the search space.

As an example, for the smallest Classic McEliece parameters we estimate that the Quantum Prange's algorithm can run with 18098 qubits, while previous works would have required at least half a million qubits.

Side channel attacks, and in particular timing attacks, are a fundamental obstacle to obtaining secure implementation of algorithms and cryptographic protocols, and have been widely researched for decades. While cryptographic definitions for the security of cryptographic systems have been well established for decades, none of these accepted definitions take into account the running time information leaked from executing the system. In this work, we give the foundation of new cryptographic definitions for cryptographic systems that take into account information about their leaked running time, focusing mainly on keyed functions such as signature and encryption schemes. Specifically,

(1) We define several cryptographic properties to express the claim that the timing information does not help an adversary to extract sensitive information, e.g. the key or the queries made. We highlight the definition of key-obliviousness, which means that an adversary cannot tell whether it received the timing of the queries with the actual key or the timing of the same queries with a random key.

(2) We present a construction of key-oblivious pseudorandom permutations on a small or medium-sized domain. This construction is not ``fixed-time,'' and at the same time is secure against any number of queries even in case the adversary knows the running time exactly. Our construction, which we call Janus Sometimes Recurse, is a variant of the ``Sometimes Recurse'' shuffle by Morris and Rogaway.

(3) We suggest a new security notion for keyed functions, called noticeable security, and prove that cryptographic schemes that have noticeable security remain secure even when the exact timings are leaked, provided the implementation is key-oblivious. We show that our notion applies to cryptographic signatures, private key encryption and PRPs.

We present an efficient quantum algorithm for solving the semidirect discrete logarithm problem (SDLP) in any finite group. The believed hardness of the semidirect discrete logarithm problem underlies more than a decade of works constructing candidate post-quantum cryptographic algorithms from nonabelian groups. We use a series of reduction results to show that it suffices to consider SDLP in finite simple groups. We then apply the celebrated Classification of Finite Simple Groups to consider each family. The infinite families of finite simple groups admit, in a fairly general setting, linear algebraic attacks providing a reduction to the classical discrete logarithm problem. For the sporadic simple groups, we show that their inherent properties render them unsuitable for cryptographically hard SDLP instances, which we illustrate via a Baby-Step Giant-Step style attack against SDLP in the Monster Group.

Our quantum SDLP algorithm is fully constructive for all but three remaining cases that appear to be gaps in the literature on constructive recognition of groups; for these cases SDLP is no harder than finding a linear representation. We conclude that SDLP is not a suitable post-quantum hardness assumption for any choice of finite group.

A popular way to build post-quantum signature schemes is by first constructing an identification scheme (IDS) and applying the Fiat-Shamir transform to it. In this work we tackle two open questions related to the general applicability of techniques around this approach that together allow for efficient post-quantum signatures with optimal security bounds in the QROM.

First we consider a recent work by Aguilar-Melchor, Hülsing, Joseph, Majenz, Ronen, and Yue (Asiacrypt'23) that showed that an optimal bound for three-round commit & open IDS by Don, Fehr, Majenz, and Schaffner (Crypto'22) can be applied to the five-round Syndrome-Decoding in the Head (SDitH) IDS. For this, they first applied a transform that replaced the first three rounds by one. They left it as an open problem if the same approach applies to other schemes beyond SDitH. We answer this question in the affirmative, generalizing their round-elimination technique and giving a generic security proof for it. Our result applies to any IDS with $2n+1$ rounds for $n>1$. However, a scheme has to be suitable for the resulting bound to not be trivial. We find that IDS are suitable when they have a certain form of special-soundness which many commit & open IDS have.

Second, we consider the hypercube technique by Aguilar-Melchor, Gama, Howe, Hülsing, Joseph, and Yue (Eurocrypt'23). An optimization that was proposed in the context of SDitH and is now used by several of the contenders in the NIST signature on-ramp. It was conjectured that the technique applies generically for the MPC-in-the-Head (MPCitH) technique that is used in the design of many post-quantum IDS if they use an additive secret sharing scheme but this was never proven. In this work we show that the technique generalizes to MPCitH IDS that use an additively homomorphic MPC protocol, and we prove that security is preserved.

We demonstrate the application of our results to the identification scheme of RYDE, a contender in the recent NIST signature on-ramp. While RYDE was already specified with the hypercube technique applied, this gives the first QROM proof for RYDE with an optimally tight bound.

Decentralized payments have evolved from using pseudonymous identifiers to much more elaborate mechanisms to ensure privacy. They can shield the amounts in payments and achieve untraceability, e.g., decoy-based untraceable payments use decoys to obfuscate the actual asset sender or asset receiver. There are two types of decoy-based payments: full decoy set payments that use all other available users as decoys, e.g., Zerocoin, Zerocash, and ZCash, and user-defined decoy set payments where the users select small decoy sets from available users, e.g., Monero, Zether, and QuisQuis.

Existing decoy-based payments face at least two of the following problems: (1) degrading untraceability due to the possibility of payment-graph analysis in user-defined decoy payments, (2) trusted setup, (3) availability issues due to expiring transactions in full decoy sets and epochs, and (4) an ever-growing set of unspent outputs since transactions keep generating outputs without saying which ones are spent. QuisQuis is the first one to solve all these problems; however, QuisQuis requires large cryptographic proofs for validity.

We introduce Nopenena (means ``cannot see''): account-based, confidential, and user-defined decoy set payment protocol, that has short proofs and also avoids these four issues. Additionally, Nopenena can be integrated with zero-knowledge contracts like Zether's $\Sigma-$Bullets and Confidential Integer Processing (CIP) to build decentralized applications. Nopenena payments are about 80% smaller than QuisQuis payments due to Nopenena's novel cryptographic protocol. Therefore, decentralized systems benefit from Nopenena's untraceability and efficiency.

The field of verifiable secret sharing schemes was introduced by Verheul et al. and has evolved over time, including well-known examples by Feldman and Pedersen. Stinson made advancements in combinatorial design-based secret sharing schemes in 2004. Desmedt et al. introduced the concept of frameproofness in 2021, while recent research by Sehrawat et al. in 2021 focuses on LWE-based access structure hiding verifiable secret sharing with malicious-majority settings. Furthermore, Roy et al. combined the concepts of reparable threshold schemes by Stinson et al. and frameproofness by Desmedt et al. in 2023, to develop extendable tensor designs built from balanced incomplete block designs, and also presented a frameproof version of their design. This paper explores ramp-type verifiable secret sharing schemes, and the application of hidden access structures in such cryptographic protocols. Inspired by Sehrawat et al.'s access structure hiding scheme, we develop an $\epsilon$-almost access structure hiding scheme, which is verifiable as well as frameproof. We detail how the concept $\epsilon$-almost hiding is important for incorporating ramp schemes, thus making a fundamental generalisation of this concept.

This note shows practical committing attacks against Rocca-S, an authenticated encryption with associated data scheme designed for 6G applications. Previously, the best complexity of the attack was $2^{64}$ by Derbez et al. in ToSC 2024(1)/FSE 2024. We show that the committing attack against Rocca by Takeuchi et al. in ToSC 2024(2)/FSE 2025 can be applied to Rocca-S, where Rocca is an earlier version of Rocca-S. We show a concrete test vector of our attack. We also point out a committing attack that exploits equivalent keys.