• 정보보호학회논문지
• /
• 제34권3호
• /
• pp.451-461
• /
• 2024

최근 거대 언어 모델을 기반으로 한 다양한 인공지능 챗봇이 출시되고 있다. 챗봇은 대화형 프롬프트를 통해 사용자에게 빠르고 간편하게 정보를 제공할 수 있다는 이점을 가지고 있어서 질의응답, 글쓰기, 프로그래밍 등 다양한 분야에서 활용되고 있다. 그러나 최근에는 챗봇의 취약점을 악용하는 '프롬프트 주입 공격'이 제안되었는데, 이는 챗봇이 기입력된 지시사항을 위반하도록 하는 공격이다. 이와 같은 공격은 거대 언어 모델 내부의 기밀 정보를 유출하거나 또 다른 악성 행위를 유발할 수 있어서 치명적이다. 반면 이들에 대한 취약점 여부가 한국어 프롬프트를 대상으로는 충분히 검증되지 않았다. 따라서 본 논문에서는 널리 사용되는 챗봇인 ChatGPT를 대상으로 악성 한국어 프롬프트를 생성하여 공격을 수행해보고, 이들에 대한 실행 가능성을 분석하고자 한다. 이를 위해 기존에 제안된 프롬프트 주입 공격 기법을 분석하여 악의적인 한국어 프롬프트를 자동으로 생성하는 시스템을 제안하고자 한다. 특히 유해 표현을 유도하는 악성 프롬프트를 중점적으로 생성하였고 이들이 실제 유효함을 보이도록 한다.

• 한국산학기술학회논문지
• /
• 제15권8호
• /
• pp.5256-5262
• /
• 2014

최근 국가기관, 언론사, 금융권 등에 대하여 분산 서비스 거부(Distributed Denial of Service, DDoS) 공격, 악성코드 유포 등 무차별 사이버테러가 발생하고 있다. DDoS 공격은 네트워크 계층에서의 대역폭 소모를 주된 공격 방법으로 정상적인 사용자와 크게 다르지 않는 패킷을 이용하여 공격을 하기 때문에 탐지 및 대응이 어렵다. 이러한 인터넷 비정상적인 트래픽이 증가하여 네트워크의 안전성 및 신뢰성을 위협하고 있어 비정상 트래픽에 대한 발생 징후를 사전에 탐지하여 대응할 수 있는 방안의 필요성이 대두되고 있다. 본 연구에서는 비정상 트래픽 탐지 기법에 대한 현황 및 문제점을 분석하고, 예측방법인 추세 모형, 지수평활법, 웨이브렛 분석 방법 등을 비교 분석하여 인터넷 트래픽의 특성을 실시간으로 분석 및 예측이 가능한 가장 적합한 예측 모형을 이용한 탐지 방법을 제안하고자 한다.

• 정보보호학회논문지
• /
• 제28권4호
• /
• pp.847-857
• /
• 2018

악성코드의 수가 가파르게 증가하면서 기업 및 공공기관, 금융기관, 병 의원 등을 타깃으로 한 사이버 공격 피해 사례가 늘어나고 있다. 이러한 흐름에 따라 학계와 보안 업계에서는 악성코드 탐지를 위한 다양한 연구를 진행하고 있다. 최근 들어서는 딥러닝을 비롯해 머신러닝 기법을 적용하는 형태의 연구가 많이 진행되는 추세다. 이 중 합성곱 신경망(CNN: Convolutional Neural Network), ResNet 등을 이용한 악성코드 분류 연구의 경우에는 기존의 분류 방법에 비해 정확도가 크게 향상된 것을 확인할 수 있다. 그러나 타깃 공격의 특징 중 하나는 사용된 악성코드가 불특정 다수를 상대로 광범위하게 퍼뜨리는 형태가 아닌, 특정 대상을 타깃으로 한 맞춤형 악성코드라는 점이다. 이러한 유형의 악성코드는 그 수가 많지 않기 때문에 기존에 연구되어온 머신러닝이나 딥러닝 기법을 적용하기에 한계가 있다. 본 논문은 타깃형 악성코드와 같이 샘플의 양이 부족한 상황에서 악성코드를 분류하는 방법에 대해 다루고 있다. 메모리가 추가된 신경망(MANN: Memory Augmented Neural Networks) 모델을 이용하였고 각 그룹별 20개의 소량 데이터로 구성되어 있는 악성코드 데이터셋에 대해 최대 97%까지 정확도로 분류할 수 있음을 확인하였다.

• Journal of Information Processing Systems
• /
• 제16권4호
• /
• pp.760-773
• /
• 2020

When a crime occurs, the information necessary for solving the case, and various pieces of the evidence needed to prove the crime are collected from the crime scene. The tangible residues collected through scientific methods at the crime scene become evidence at trial and a clue to prove the facts directly against the offense of the suspect. Therefore, the scientific investigation and forensic handling for securing objective forensic in crime investigation is increasingly important. Today, digital systems, such as smartphones, CCTVs, black boxes, etc. are increasingly used as criminal information investigation clues, and digital forensic is becoming a decisive factor in investigation and trial. However, the systems have the risk that digital forensic may be damaged or manipulated by malicious insiders in the existing centralized management systems based on client/server structure. In this paper, we design and implement a blockchain based digital forensic management model using Hyperledger Fabric and Docker to guarantee the reliability and integrity of digital forensic. The proposed digital evidence management model allows only authorized participants in a distributed environment without a central management agency access the network to share and manage potential crime data. Therefore, it could be relatively safe from malicious internal attackers compared to the existing client/server model.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권7호
• /
• pp.3386-3411
• /
• 2019

Secured and reliable routing is a crucial factor for improving the performance of Wireless Mesh Networks (WMN) since these networks are susceptible to many types of attacks. The existing assumption about the internal nodes in wireless mesh networks is that they cooperate well during the forwarding of packets all the time. However, it is not always true due to the presence of malicious and mistrustful nodes. Hence, it is essential to establish a secure, reliable and stable route between a source node and a destination node in WMN. In this paper, a trust based secure routing algorithm is proposed for enhancing security and reliability of WMN, which contains cross layer and subject logic based reliable reputation scheme with security tag model for providing effective secured routing. This model uses only the trusted nodes with the forwarding reliability of data transmission and it isolates the malicious nodes from the providing path. Moreover, every node in this model is assigned with a security tag that is used for efficient authentication. Thus, by combining authentication, trust and subject logic, the proposed approach is capable of choosing the trusted nodes effectively to participate in forwarding the packets of trustful peer nodes successfully. The simulation results obtained from this work show that the proposed routing protocol provides optimal network performance in terms of security and packet delivery ratio.

• Journal of Information Processing Systems
• /
• 제18권5호
• /
• pp.650-664
• /
• 2022

Mobile applications can be easily downloaded and installed via markets. However, malware and malicious applications containing unwanted advertisements exist in these application markets. Therefore, smartphone users install applications with reference to the application review to avoid such malicious applications. An application review typically comprises contents for evaluation; however, a false review with a specific purpose can be included. Such false reviews are known as fake reviews, and they can be generated using artificial intelligence (AI)-based text-generating models. Recently, AI-based text-generating models have been developed rapidly and demonstrate high-quality generated texts. Herein, we analyze the features of fake reviews generated from Generative Pre-Training-2 (GPT-2), an AI-based text-generating model and create a model to detect those fake reviews. First, we collect a real human-written application review from Kaggle. Subsequently, we identify features of the fake review using natural language processing and statistical analysis. Next, we generate fake review detection models using five types of machine-learning models trained using identified features. In terms of the performances of the fake review detection models, we achieved average F1-scores of 0.738, 0.723, and 0.730 for the fake review, real review, and overall classifications, respectively.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권3호
• /
• pp.1502-1522
• /
• 2019

Leakage of secret keys may be the most devastating problem in public key cryptosystems because it means that all security guarantees are missing. The forward security mechanism allows users to update secret keys frequently without updating public keys. Meanwhile, it ensures that an attacker is unable to derive a user's secret keys for any past time, even if it compromises the user's current secret key. Therefore, it offers an effective cryptographic approach to address the private key leakage problem. As an extension of the forward security mechanism in certificate-based public key cryptography, forward-secure certificate-based signature (FS-CBS) has many appealing merits, such as no key escrow, no secure channel and implicit authentication. Until now, there is only one FS-CBS scheme that does not employ the random oracles. Unfortunately, our cryptanalysis indicates that the scheme is subject to the security vulnerability due to the existential forgery attack from the malicious CA. Our attack demonstrates that a CA can destroy its existential unforgeability by implanting trapdoors in system parameters without knowing the target user's secret key. Therefore, it is fair to say that to design a FS-CBS scheme secure against malicious CAs without lying random oracles is still an unsolved issue. To address this problem, we put forward an enhanced FS-CBS scheme without random oracles. Our FS-CBS scheme not only fixes the security weakness in the original scheme, but also significantly optimizes the scheme efficiency. In the standard model, we formally prove its security under the complexity assumption of the square computational Diffie-Hellman problem. In addition, the comparison with the original FS-CBS scheme shows that our scheme offers stronger security guarantee and enjoys better performance.

• ETRI Journal
• /
• 제39권3호
• /
• pp.406-416
• /
• 2017

Malware propagated via the World Wide Web is one of the most dangerous tools in the realm of cyber-attacks. Its methodologies are effective, relatively easy to use, and are developing constantly in an unexpected manner. As a result, rapidly detecting malware propagation websites from a myriad of webpages is a difficult task. In this paper, we present LoGos, an automated high-interaction dynamic analyzer optimized for a browser-based Windows virtual machine environment. LoGos utilizes Internet Explorer injection and API hooks, and scrutinizes malicious behaviors such as new network connections, unused open ports, registry modifications, and file creation. Based on the obtained results, LoGos can determine the maliciousness level. This model forms a very lightweight system. Thus, it is approximately 10 to 18 times faster than systems proposed in previous work. In addition, it provides high detection rates that are equal to those of state-of-the-art tools. LoGos is a closed tool that can detect an extensive array of malicious webpages. We prove the efficiency and effectiveness of the tool by analyzing almost 0.36 M domains and 3.2 M webpages on a daily basis.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권10호
• /
• pp.3793-3814
• /
• 2021

Most secure solutions like cryptography are software based and they are designed to mainly deal with the outside attacks for traditional networks, but such soft security is hard to be implemented in wireless sensor networks to counter the inside attacks from internal malicious nodes. To address this issue, reputation has been introduced to tackle the inside malicious nodes. Reputation is essentially a stimulating mechanism for nodes' cooperation and is employed to detect node misbehaviors and improve the trust-worthiness between individual nodes. Among the reputation models, binomial distribution based reputation has many advantages such as light weight and ease of implementation in resource-constraint sensor nodes, and accordingly researchers have proposed many insightful related methods. However, some of them either directly use the modelling results, apply the models through simple modifications, or only use the required components while ignoring the others as an integral part of the whole model, this topic still lacks a comprehensive and systematical review. Thus the motivation of this study is to provide a thorough survey concerning each detailed functional components of binomial distribution based reputation for wireless sensor networks. In addition, based on the survey results, we also argue some open research problems and suggest the directions that are worth future efforts. We believe that this study is helpful to better understanding the reputation modeling mechanism and its components for wireless sensor networks, and can further attract more related future studies.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권11호
• /
• pp.4224-4243
• /
• 2021

Cognitive radio (CR) is a feasible intelligent technology and can be used as an effective solution to spectrum scarcity and underutilization. As the key function of CR, cooperative spectrum sensing (CSS) is able to effectively prevent the harmful interference with primary users (PUs) and identify the available spectrum resources by exploiting the spatial diversity of multiple secondary users (SUs). However, the open nature of the cognitive radio networks (CRNs) framework makes CSS face many security threats, such as, the malicious user (MU) launches Byzantine attack to undermine CRNs. For this aim, we make an in-depth analysis of the motive and purpose from the MU's perspective in the interweave CR system, aiming to provide the future guideline for defense strategies. First, we formulate a dynamic Byzantine attack model by analyzing Byzantine behaviors in the process of CSS. On the basis of this, we further make an investigation on the condition of making the fusion center (FC) blind when the fusion rule is unknown for the MU. Moreover, the throughput and interference to the primary network are taken into consideration to evaluate the impact of Byzantine attack on the interweave CR system, and then analyze the optimal strategy of Byzantine attack when the fusion rule is known. Finally, theoretical proofs and simulation results verify the correctness and effectiveness of analyses about the impact of Byzantine attack strategy on the throughput and interference.