Large Vision-Language Models (LVLMs) signify a groundbreaking paradigm shift within the Artificial Intelligence (AI) community, extending beyond the capabilities of Large Language Models (LLMs) by assimilating additional modalities (e.g., images). Despite this advancement, the safety of LVLMs remains adequately underexplored, with a potential overreliance on the safety assurances purported by their underlying LLMs. In this paper, we propose FigStep, a straightforward yet effective black-box jailbreak algorithm against LVLMs. Instead of feeding textual harmful instructions directly, FigStep converts the prohibited content into images through typography to bypass the safety alignment. The experimental results indicate that FigStep can achieve an average attack success rate of 82.50% on six promising open-source LVLMs. Not merely to demonstrate the efficacy of FigStep, we conduct comprehensive ablation studies and analyze the distribution of the semantic embeddings to uncover that the reason behind the success of FigStep is the deficiency of safety alignment for visual embeddings. Moreover, we compare FigStep with five text-only jailbreaks and four image-based jailbreaks to demonstrate the superiority of FigStep, i.e., negligible attack costs and better attack performance. Above all, our work reveals that current LVLMs are vulnerable to jailbreak attacks, which highlights the necessity of novel cross-modality safety alignment techniques.

Private set union (PSU) is a cryptographic protocol that allows two parties to compute the union of their sets without revealing anything else. Despite some efficient PSU protocols that have been proposed, they mainly focus on the balanced setting, where the sets held by the parties are of similar size. Recently, Tu et al. (CCS 2023) proposed the first unbalanced PSU protocol which achieves sublinear communication complexity in the size of the larger set.

Abstract The current widely used public‐key cryptosystems are vulnerable to quantum attacks. To prepare for cybersecurity in the quantum era, some projects have been launched to call for post‐quantum alternatives. Due to solid security and desirable performance, lattice‐based cryptosystems are viewed as promising candidates in the upcoming standardisation of post‐quantum cryptography. This study surveys the lattice‐based cryptosystems in the post‐quantum standardisation processes including the NIST Post‐Quantum Cryptography Standardisation and the Chinese Cryptographic Algorithm Design Competition, from both design and security aspects. We present generic design paradigms of lattice‐based schemes and describe several representative proposals and recent progress. We also recap some main cryptanalytic results and methods for estimating the concrete security of lattice‐based schemes.

Abstract NewHope cryptosystem is one of the second-round submissions of the National Institute of Standards and Technology post-quantum cryptography standardization process, which is a suite of two key encapsulation mechanisms based on the ring-learning with errors (LWE) problem. It has received much attention from the research community due to its small key size and high efficiency. Recently, three key mismatch attacks are proposed against NewHope under the condition of key reuse. They do not solve the ring-LWE instance directly but exploit the leakage of secret information. As far as we know, the best result is given by Okada et al. ((2020) Improving Key Mismatch Attack on NewHope with Fewer Queries. In Proc. of the 25th Australasian Conf. on Information Security and Privacy, Perth, WA, Australia, November 30–December 2, pp. 505–524. Springer Cham, Switzerland), which recovers the whole secret with a success probability of $97\%$ and $233,803$ average queries. In this paper, we further improve the key mismatch attack of NewHope by reducing the average queries to $106,577$ and raising the success probability to $100\%$. Moreover, we analyze the key mismatch attack without key reuse for the first time and we propose a combinatorial attack against NewHope1024. The total complexity of the combinatorial attack is $2^{253}$, which is lower than the complexity of primal attack and the claimed security strength of NewHope1024.

For locally repairable codes (LRCs), Cadambe and Mazumdar derived the first field-dependent parameter bound, known as the C-M bound. However, the C-M bound depends on an undetermined parameter kopt(q)(n, d). In this paper, a sphere-packing approach is developed for upper bounding the parameter k for [n, k, d] linear LRCs with locality r. When restricted to the binary field, three upper bounds (i.e., Bound A, Bound B, and Bound C) are derived in an explicit form. More specifically, Bound A holds under the hypothesis that the local repair groups are disjoint and of equal size. Comparing with previous bounds obtained under the same hypothesis, Bound A either covers them as special cases or has an advantage due to its explicit form. Then, the hypothesis is removed in Bound B and Bound C. As the price for explicit form, Bound B specially holds for d ≥ 5 and Bound C for r = 2. Through specific comparisons, we show that Bound B and Bound C both tend to outperform the C-M bound, as n goes large. Moreover, a family of binary linear LRCs with d ≥ 6 attaining Bound B are constructed and later extended to a wider range of parameters by a shortening technique. Lastly, most of the bounds and constructions are extended to q-ary LRCs.

For binary [n, k, d] linear locally repairable codes (LRCs), two new upper bounds on k are derived. The first one applies to LRCs with disjoint local repair groups, for general values of n, d and locality r, containing some previously known bounds as special cases. The second one is based on solving an optimization problem and applies to LRCs with arbitrary structure of local repair groups. Particularly, an explicit bound is derived from the second bound when d ≥ 5. A specific comparison shows this explicit bound outperforms the Cadambe-Mazumdar bound for 5 ≤ d ≤ 8 and large values of n. Moreover, a construction of binary linear LRCs with d ≥ 6 attaining our second bound is provided.

An (r, t)-locally repairable code satisfies a property that the value at each coordinate can be recovered from t disjoint repair sets each containing at most r other coordinates. This property is extremely useful in distributed storage systems for hot data. In this paper, we propose two constructions of (r, t)-LRCs. The first one is a cyclic code of which the parity check polynomial is closely related to the trace function over finite fields. This code can achieve high availability and large minimum distance. The second one is based on the inclusion matrix of linear subspaces in Fqm. For some specific parameters, we prove that its information rate is always higher than r over r+t which was conjectured to be near to the optimal rate for (r, t)-LRCs (A. Wang and Z. Zhang, ISIT 2015). By shortening this code in a specially designed way, we obtain (r, t)-LRCs with more desirable locality r at a slight expense of information rate.

The ith coordinate of an [n, k] code is said to have locality r and availability t if there exist t disjoint groups, each containing at most r other coordinates that can together recover the value of the ith coordinate. This property is particularly useful for codes for distributed storage systems because it permits local repair of failed nodes and parallel access of hot data. In this paper, for any positive integers r and t, we construct a binary linear code of length equation which has locality r and availability t for all coordinates. Although it only achieves the trivial minimum distance (i.e. t + 1), its information rate attains equation, which is higher than that of the direct product code, the only known construction that can achieve arbitrary locality and availability.

The locally repairable code (LRC) studied in this paper is an [n, k] linear code of which the value at each coordinate can be recovered by a linear combination of at most r other coordinates. The central problem in this paper is to determine the largest possible minimum distance for LRCs. First, an integer programming-based upper bound is derived for any LRC. Then, by solving the programming problem under certain conditions, an explicit upper bound is obtained for LRCs with parameters n1> n2, where n1= ⌈(n/r + 1)⌉ and n2 = n1(r +1)-n. Finally, an explicit construction for LRCs attaining this upper bound is presented over the finite field F2m,where m ≥ n1r. Based on these r ≤ √n - 1 has been definitely determined, which is of great results, the largest possible minimum distance for all LRCs with significance in practical use.

Repair locality is a desirable property for erasure codes in distributed storage systems. Recently, different structures of local repair groups have been proposed in the definitions of repair locality. In this paper, the concept of regenerating set is introduced to characterize the local repair groups. A definition of locality r(δ-1)(i.e., locality r with repair tolerance δ - 1) under the most general structure of regenerating sets is given. All previously studied locality notions turn out to be special cases of this definition. Furthermore, three representative notions of locality proposed before are reinvestigated under the framework of regenerating sets, and their respective upper bounds on the minimum distance are reproved in a uniform and brief form. Additionally, a tighter distance bound is derived for the square code which is a class of linear codes with locality r(2)and high information rate, and an explicit code construction attaining the optimal distance bound is obtained.

In distributed storage systems, erasure codes with locality r are preferred because a coordinate can be locally repaired by accessing at most r other coordinates which in turn greatly reduces the disk I/O complexity for small r. However, the local repair may not be performed when some of the r coordinates are also erased. To overcome this problem, we propose the (r, δ)c-locality providing δ-1 nonoverlapping local repair groups of size no more than r for a coordinate. Consequently, the repair locality r can tolerate δ -1 erasures in total. We derive an upper bound on the minimum distance for any linear [n, k] code with information (r, δ)c-locality. Then, we prove existence of the codes that attain this bound when n ≥ k(r(δ - 1) + 1). Although the locality (r, δ) defined by Prakash et al. provides the same level of locality and local repair tolerance as our definition, codes with (r, δ)c-locality attaining the bound are proved to have more advantage in the minimum distance. In particular, we construct a class of codes with all symbol (r, δ)c-locality where the gain in minimum distance is Q(√r) and the information rate is close to 1.

We give an explicit construction of exact cooperative regenerating codes at the MBCR (minimum bandwidth cooperative regeneration) point. Before the paper, the only known explicit MBCR codes are given with parameters n = d + r and d = k, while our construction applies to all possible values of n, k, d, r. The code has a brief expression in the polynomial form and the data reconstruction is accomplished by bivariate polynomial interpolation. It is a scalar code and operates over a finite field of size q ≥ n. Besides, we establish several subspace properties for linear exact MBCR codes. Based on these properties we prove that linear exact MBCR codes cannot achieve repair-by-transfer.