Pub Date : 2023-05-01DOI: 10.1109/SP46215.2023.10179298
Yu Hao, Guoren Li, Xiaochen Zou, Weiteng Chen, Shitong Zhu, Zhiyun Qian, A. A. Sani
Fuzz testing operating system kernels has been effective overall in recent years. For example, syzkaller manages to find thousands of bugs in the Linux kernel since 2017. One necessary component of syzkaller is a collection of syscall descriptions that are often provided by human experts. However, to our knowledge, current syscall descriptions are largely written manually, which is both time-consuming and error-prone. It is especially challenging considering that there are many kernel drivers (for new hardware devices and beyond) that are continuously being developed and evolving over time. In this paper, we present a principled solution for generating syscall descriptions for Linux kernel drivers. At its core, we summarize and model the key invariants or programming conventions, extracted from the "contract" between the core kernel and drivers. This allows us to understand programmatically how a kernel driver is initialized and how its associated interfaces are constructed. With this insight, we have developed a solution in a tool called SyzDescribe that has been tested for over hundreds of kernel drivers. We show that the syscall descriptions produced by SyzDescribe are competitive to manually-curated ones, and much better than prior work (i.e., DIFUZE and KSG). Finally, we analyze the gap between our descriptions and the ground truth and point to future improvement opportunities.
{"title":"SyzDescribe: Principled, Automated, Static Generation of Syscall Descriptions for Kernel Drivers","authors":"Yu Hao, Guoren Li, Xiaochen Zou, Weiteng Chen, Shitong Zhu, Zhiyun Qian, A. A. Sani","doi":"10.1109/SP46215.2023.10179298","DOIUrl":"https://doi.org/10.1109/SP46215.2023.10179298","url":null,"abstract":"Fuzz testing operating system kernels has been effective overall in recent years. For example, syzkaller manages to find thousands of bugs in the Linux kernel since 2017. One necessary component of syzkaller is a collection of syscall descriptions that are often provided by human experts. However, to our knowledge, current syscall descriptions are largely written manually, which is both time-consuming and error-prone. It is especially challenging considering that there are many kernel drivers (for new hardware devices and beyond) that are continuously being developed and evolving over time. In this paper, we present a principled solution for generating syscall descriptions for Linux kernel drivers. At its core, we summarize and model the key invariants or programming conventions, extracted from the \"contract\" between the core kernel and drivers. This allows us to understand programmatically how a kernel driver is initialized and how its associated interfaces are constructed. With this insight, we have developed a solution in a tool called SyzDescribe that has been tested for over hundreds of kernel drivers. We show that the syscall descriptions produced by SyzDescribe are competitive to manually-curated ones, and much better than prior work (i.e., DIFUZE and KSG). Finally, we analyze the gap between our descriptions and the ground truth and point to future improvement opportunities.","PeriodicalId":439989,"journal":{"name":"2023 IEEE Symposium on Security and Privacy (SP)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116601748","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/SP46215.2023.10179353
Norbert Ludant, Pieter Robyns, G. Noubir
We present the first open-source tool capable of efficiently sniffing 5G control channels, 5GSniffer and demonstrate its potential to conduct attacks on users privacy. 5GSniffer builds on our analysis of the 5G RAN control channel exposing side-channel leakage. We note that decoding the 5G control channels is significantly more challenging than in LTE, since part of the information necessary for decoding is provided to the UEs over encrypted channels. We devise a set of techniques to achieve real-time control channels sniffing (over three orders of magnitude faster than brute-forcing). This enables, among other things, to retrieve the Radio Network Temporary Identifiers (RNTIs) of all users in a cell, and perform traffic analysis. To illustrate the potential of our sniffer, we analyse two privacy-focused messengers, Signal and Telegram. We identify privacy leaks that can be exploited to generate stealthy traffic to a target user. When combined with 5GSniffer, it enables stealthy exposure of the presence of a target user in a given location (solely based on their phone number), by linking the phone number to the RNTI. It also enables traffic analysis of the target user. We evaluate the attacks and our sniffer, demonstrating nearly 100% accuracy within 30 seconds of attack initiation.
{"title":"From 5G Sniffing to Harvesting Leakages of Privacy-Preserving Messengers","authors":"Norbert Ludant, Pieter Robyns, G. Noubir","doi":"10.1109/SP46215.2023.10179353","DOIUrl":"https://doi.org/10.1109/SP46215.2023.10179353","url":null,"abstract":"We present the first open-source tool capable of efficiently sniffing 5G control channels, 5GSniffer and demonstrate its potential to conduct attacks on users privacy. 5GSniffer builds on our analysis of the 5G RAN control channel exposing side-channel leakage. We note that decoding the 5G control channels is significantly more challenging than in LTE, since part of the information necessary for decoding is provided to the UEs over encrypted channels. We devise a set of techniques to achieve real-time control channels sniffing (over three orders of magnitude faster than brute-forcing). This enables, among other things, to retrieve the Radio Network Temporary Identifiers (RNTIs) of all users in a cell, and perform traffic analysis. To illustrate the potential of our sniffer, we analyse two privacy-focused messengers, Signal and Telegram. We identify privacy leaks that can be exploited to generate stealthy traffic to a target user. When combined with 5GSniffer, it enables stealthy exposure of the presence of a target user in a given location (solely based on their phone number), by linking the phone number to the RNTI. It also enables traffic analysis of the target user. We evaluate the attacks and our sniffer, demonstrating nearly 100% accuracy within 30 seconds of attack initiation.","PeriodicalId":439989,"journal":{"name":"2023 IEEE Symposium on Security and Privacy (SP)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116677676","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/SP46215.2023.10179331
W. Zong, Yang-Wai Chow, Willy Susilo, Kien Do, S. Venkatesh
While deep learning techniques have achieved great success in modern digital products, researchers have shown that deep learning models are susceptible to Trojan attacks. In a Trojan attack, an adversary stealthily modifies a deep learning model such that the model will output a predefined label whenever a trigger is present in the input. In this paper, we present TrojanModel, a practical Trojan attack against Automatic Speech Recognition (ASR) systems. ASR systems aim to transcribe voice input into text, which is easier for subsequent downstream applications to process. We consider a practical attack scenario in which an adversary inserts a Trojan into the acoustic model of a target ASR system. Unlike existing work that uses noise-like triggers that will easily arouse user suspicion, the work in this paper focuses on the use of unsuspicious sounds as a trigger, e.g., a piece of music playing in the background. In addition, TrojanModel does not require the retraining of a target model. Experimental results show that TrojanModel can achieve high attack success rates with negligible effect on the target model’s performance. We also demonstrate that the attack is effective in an over-the-air attack scenario, where audio is played over a physical speaker and received by a microphone.
{"title":"TrojanModel: A Practical Trojan Attack against Automatic Speech Recognition Systems","authors":"W. Zong, Yang-Wai Chow, Willy Susilo, Kien Do, S. Venkatesh","doi":"10.1109/SP46215.2023.10179331","DOIUrl":"https://doi.org/10.1109/SP46215.2023.10179331","url":null,"abstract":"While deep learning techniques have achieved great success in modern digital products, researchers have shown that deep learning models are susceptible to Trojan attacks. In a Trojan attack, an adversary stealthily modifies a deep learning model such that the model will output a predefined label whenever a trigger is present in the input. In this paper, we present TrojanModel, a practical Trojan attack against Automatic Speech Recognition (ASR) systems. ASR systems aim to transcribe voice input into text, which is easier for subsequent downstream applications to process. We consider a practical attack scenario in which an adversary inserts a Trojan into the acoustic model of a target ASR system. Unlike existing work that uses noise-like triggers that will easily arouse user suspicion, the work in this paper focuses on the use of unsuspicious sounds as a trigger, e.g., a piece of music playing in the background. In addition, TrojanModel does not require the retraining of a target model. Experimental results show that TrojanModel can achieve high attack success rates with negligible effect on the target model’s performance. We also demonstrate that the attack is effective in an over-the-air attack scenario, where audio is played over a physical speaker and received by a microphone.","PeriodicalId":439989,"journal":{"name":"2023 IEEE Symposium on Security and Privacy (SP)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122493340","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/SP46215.2023.10179308
Maximilian Noppel, Lukas Peter, Christian Wressnegger
Explainable machine learning holds great potential for analyzing and understanding learning-based systems. These methods can, however, be manipulated to present unfaithful explanations, giving rise to powerful and stealthy adversaries. In this paper, we demonstrate how to fully disguise the adversarial operation of a machine learning model. Similar to neural backdoors, we change the model’s prediction upon trigger presence but simultaneously fool an explanation method that is applied post-hoc for analysis. This enables an adversary to hide the presence of the trigger or point the explanation to entirely different portions of the input, throwing a red herring. We analyze different manifestations of these explanation-aware backdoors for gradient- and propagation-based explanation methods in the image domain, before we resume to conduct a red-herring attack against malware classification.
{"title":"Disguising Attacks with Explanation-Aware Backdoors","authors":"Maximilian Noppel, Lukas Peter, Christian Wressnegger","doi":"10.1109/SP46215.2023.10179308","DOIUrl":"https://doi.org/10.1109/SP46215.2023.10179308","url":null,"abstract":"Explainable machine learning holds great potential for analyzing and understanding learning-based systems. These methods can, however, be manipulated to present unfaithful explanations, giving rise to powerful and stealthy adversaries. In this paper, we demonstrate how to fully disguise the adversarial operation of a machine learning model. Similar to neural backdoors, we change the model’s prediction upon trigger presence but simultaneously fool an explanation method that is applied post-hoc for analysis. This enables an adversary to hide the presence of the trigger or point the explanation to entirely different portions of the input, throwing a red herring. We analyze different manifestations of these explanation-aware backdoors for gradient- and propagation-based explanation methods in the image domain, before we resume to conduct a red-herring attack against malware classification.","PeriodicalId":439989,"journal":{"name":"2023 IEEE Symposium on Security and Privacy (SP)","volume":"150 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115168587","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/SP46215.2023.10179474
Xiaohan Zhang, Haoqi Ye, Ziqi Huang, Xiao Ye, Yinzhi Cao, Yuan Zhang, Min Yang
Face Verification Systems (FVSes) are more and more deployed by real-world mobile applications (apps) to verify a human’s claimed identity. One popular type of FVSes is called cross-side FVS (XFVS), which splits the FVS functionality into two sides: one at a mobile phone to take pictures or videos and the other at a trusted server for verification. Prior works have studied the security of XFVSes from the machine learning perspective, i.e., whether the learning models used by XFVSes are robust to adversarial attacks. However, the security of other parts of XFVSes, especially the design and implementation of the verification procedure used by XFVSes, is not well understood.In this paper, we conduct the first measurement study on the security of real-world XFVSes used by popular mobile apps from a system perspective. More specifically, we design and implement a semi-automated system, called XFVSChecker, to detect XFVSes in mobile apps and then inspect their compliance with four security properties. Our evaluation reveals that most of existing XFVS apps, including those with billions of downloads, are vulnerable to at least one of four types of attacks. These attacks require only easily available attack prerequisites, such as one photo of the victim, to pose significant security risks, including complete account takeover, identity fraud and financial loss. Our findings result in 14 Chinese National Vulnerability Database (CNVD) IDs and one of them, particularly CNVD-2021-86899, is awarded the most valuable vulnerability in 2021 among all the reported vulnerabilities to CNVD.
{"title":"Understanding the (In)Security of Cross-side Face Verification Systems in Mobile Apps: A System Perspective","authors":"Xiaohan Zhang, Haoqi Ye, Ziqi Huang, Xiao Ye, Yinzhi Cao, Yuan Zhang, Min Yang","doi":"10.1109/SP46215.2023.10179474","DOIUrl":"https://doi.org/10.1109/SP46215.2023.10179474","url":null,"abstract":"Face Verification Systems (FVSes) are more and more deployed by real-world mobile applications (apps) to verify a human’s claimed identity. One popular type of FVSes is called cross-side FVS (XFVS), which splits the FVS functionality into two sides: one at a mobile phone to take pictures or videos and the other at a trusted server for verification. Prior works have studied the security of XFVSes from the machine learning perspective, i.e., whether the learning models used by XFVSes are robust to adversarial attacks. However, the security of other parts of XFVSes, especially the design and implementation of the verification procedure used by XFVSes, is not well understood.In this paper, we conduct the first measurement study on the security of real-world XFVSes used by popular mobile apps from a system perspective. More specifically, we design and implement a semi-automated system, called XFVSChecker, to detect XFVSes in mobile apps and then inspect their compliance with four security properties. Our evaluation reveals that most of existing XFVS apps, including those with billions of downloads, are vulnerable to at least one of four types of attacks. These attacks require only easily available attack prerequisites, such as one photo of the victim, to pose significant security risks, including complete account takeover, identity fraud and financial loss. Our findings result in 14 Chinese National Vulnerability Database (CNVD) IDs and one of them, particularly CNVD-2021-86899, is awarded the most valuable vulnerability in 2021 among all the reported vulnerabilities to CNVD.","PeriodicalId":439989,"journal":{"name":"2023 IEEE Symposium on Security and Privacy (SP)","volume":"122 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128006114","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/SP46215.2023.10179393
Kunming Jiang, Devora Chait-Roth, Zachary Destefano, Michael Walfish, Thomas Wies
There has been intense interest over the last decade in implementations of probabilistic proofs (IPs, SNARKs, PCPs, and so on): protocols in which an untrusted party proves to a verifier that a given computation was executed properly, possibly in zero knowledge. Nevertheless, implementations still do not scale beyond small computations. A central source of overhead is the front-end: translating from the abstract computation to a set of equivalent arithmetic constraints. This paper introduces a general-purpose framework, called Distiller, in which a user translates to constraints not the original computation but an abstracted specification of it. Distiller is the first in this area to perform such transformations in a way that is provably safe. Furthermore, by taking the idea of "encode a check in the constraints" to its literal logical extreme, Distiller exposes many new opportunities for constraint reduction, resulting in cost reductions for benchmark computations of 1.3–50×, and in some cases, better asymptotics.
{"title":"Less is more: refinement proofs for probabilistic proofs","authors":"Kunming Jiang, Devora Chait-Roth, Zachary Destefano, Michael Walfish, Thomas Wies","doi":"10.1109/SP46215.2023.10179393","DOIUrl":"https://doi.org/10.1109/SP46215.2023.10179393","url":null,"abstract":"There has been intense interest over the last decade in implementations of probabilistic proofs (IPs, SNARKs, PCPs, and so on): protocols in which an untrusted party proves to a verifier that a given computation was executed properly, possibly in zero knowledge. Nevertheless, implementations still do not scale beyond small computations. A central source of overhead is the front-end: translating from the abstract computation to a set of equivalent arithmetic constraints. This paper introduces a general-purpose framework, called Distiller, in which a user translates to constraints not the original computation but an abstracted specification of it. Distiller is the first in this area to perform such transformations in a way that is provably safe. Furthermore, by taking the idea of \"encode a check in the constraints\" to its literal logical extreme, Distiller exposes many new opportunities for constraint reduction, resulting in cost reductions for benchmark computations of 1.3–50×, and in some cases, better asymptotics.","PeriodicalId":439989,"journal":{"name":"2023 IEEE Symposium on Security and Privacy (SP)","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134313650","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/SP46215.2023.10179287
Jinyang Ding, Kejiang Chen, Yaofei Wang, Na Zhao, Weiming Zhang, Neng H. Yu
Steganography is the act of disguising the transmission of secret information as seemingly innocent. Although provably secure steganography has been proposed for decades, it has not been mainstream in this field because its strict requirements (such as a perfect sampler and an explicit data distribution) are challenging to satisfy in traditional data environments. The popularity of deep generative models is gradually increasing and can provide an excellent opportunity to solve this problem. Several methods attempting to achieve provably secure steganography based on deep generative models have been proposed in recent years. However, they cannot achieve the expected security in practice due to unrealistic conditions, such as the balanced grouping of discrete elements and a perfect match between the message and channel distributions. In this paper, we propose a new provably secure steganography method in practice named Discop, which constructs several "distribution copies" during the generation process. At each time step of generation, the message determines from which "distribution copy" to sample. As long as the receiver agrees on some shared information with the sender, he can extract the message without error. To further improve the embedding rate, we recursively construct more "distribution copies" by creating Huffman trees. We prove that Discop can strictly maintain the original distribution so that the adversary cannot perform better than random guessing. Moreover, we conduct experiments on multiple generation tasks for diverse digital media, and the results show that Discop’s security and efficiency outperform those of previous methods.
{"title":"Discop: Provably Secure Steganography in Practice Based on \"Distribution Copies\"","authors":"Jinyang Ding, Kejiang Chen, Yaofei Wang, Na Zhao, Weiming Zhang, Neng H. Yu","doi":"10.1109/SP46215.2023.10179287","DOIUrl":"https://doi.org/10.1109/SP46215.2023.10179287","url":null,"abstract":"Steganography is the act of disguising the transmission of secret information as seemingly innocent. Although provably secure steganography has been proposed for decades, it has not been mainstream in this field because its strict requirements (such as a perfect sampler and an explicit data distribution) are challenging to satisfy in traditional data environments. The popularity of deep generative models is gradually increasing and can provide an excellent opportunity to solve this problem. Several methods attempting to achieve provably secure steganography based on deep generative models have been proposed in recent years. However, they cannot achieve the expected security in practice due to unrealistic conditions, such as the balanced grouping of discrete elements and a perfect match between the message and channel distributions. In this paper, we propose a new provably secure steganography method in practice named Discop, which constructs several \"distribution copies\" during the generation process. At each time step of generation, the message determines from which \"distribution copy\" to sample. As long as the receiver agrees on some shared information with the sender, he can extract the message without error. To further improve the embedding rate, we recursively construct more \"distribution copies\" by creating Huffman trees. We prove that Discop can strictly maintain the original distribution so that the adversary cannot perform better than random guessing. Moreover, we conduct experiments on multiple generation tasks for diverse digital media, and the results show that Discop’s security and efficiency outperform those of previous methods.","PeriodicalId":439989,"journal":{"name":"2023 IEEE Symposium on Security and Privacy (SP)","volume":"239 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122933380","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/SP46215.2023.10179419
Kevin Choi, A. Manoj, Joseph Bonneau
Motivated and inspired by the emergence of blockchains, many new protocols have recently been proposed for generating publicly verifiable randomness in a distributed yet secure fashion. These protocols work under different setups and assumptions, use various cryptographic tools, and entail unique trade-offs and characteristics. In this paper, we systematize the design of distributed randomness beacons (DRBs) as well as the cryptographic building blocks they rely on. We evaluate protocols on two key security properties, unbiasability and unpredictability, and discuss common attack vectors for predicting or biasing the beacon output and the countermeasures employed by protocols. We also compare protocols by communication and computational efficiency. Finally, we provide insights on the applicability of different protocols in various deployment scenarios and highlight possible directions for further research.
{"title":"SoK: Distributed Randomness Beacons","authors":"Kevin Choi, A. Manoj, Joseph Bonneau","doi":"10.1109/SP46215.2023.10179419","DOIUrl":"https://doi.org/10.1109/SP46215.2023.10179419","url":null,"abstract":"Motivated and inspired by the emergence of blockchains, many new protocols have recently been proposed for generating publicly verifiable randomness in a distributed yet secure fashion. These protocols work under different setups and assumptions, use various cryptographic tools, and entail unique trade-offs and characteristics. In this paper, we systematize the design of distributed randomness beacons (DRBs) as well as the cryptographic building blocks they rely on. We evaluate protocols on two key security properties, unbiasability and unpredictability, and discuss common attack vectors for predicting or biasing the beacon output and the countermeasures employed by protocols. We also compare protocols by communication and computational efficiency. Finally, we provide insights on the applicability of different protocols in various deployment scenarios and highlight possible directions for further research.","PeriodicalId":439989,"journal":{"name":"2023 IEEE Symposium on Security and Privacy (SP)","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133792929","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/SP46215.2023.10179317
Erik Trickel, Fabio Pagani, Chang Zhu, Lukas Dresel, G. Vigna, Christopher Kruegel, Ruoyu Wang, Tiffany Bao, Yan Shoshitaishvili, Adam Doupé
Black-box web application vulnerability scanners attempt to automatically identify vulnerabilities in web applications without access to the source code. However, they do so by using a manually curated list of vulnerability-inducing inputs, which significantly reduces the ability of a black-box scanner to explore the web application’s input space and which can cause false negatives. In addition, black-box scanners must attempt to infer that a vulnerability was triggered, which causes false positives.To overcome these limitations, we propose Witcher, a novel web vulnerability discovery framework that is inspired by grey-box coverage-guided fuzzing. Witcher implements the concept of fault escalation to detect both SQL and command injection vulnerabilities. Additionally, Witcher captures coverage information and creates output-derived input guidance to focus the input generation and, therefore, to increase the state-space exploration of the web application. On a dataset of 18 web applications written in PHP, Python, Node.js, Java, Ruby, and C, 13 of which had known vulnerabilities, Witcher was able to find 23 of the 36 known vulnerabilities (64%), and additionally found 67 previously unknown vulnerabilities, 4 of which received CVE numbers. In our experiments, Witcher outperformed state of the art scanners both in terms of number of vulnerabilities found, but also in terms of coverage of web applications.
{"title":"Toss a Fault to Your Witcher: Applying Grey-box Coverage-Guided Mutational Fuzzing to Detect SQL and Command Injection Vulnerabilities","authors":"Erik Trickel, Fabio Pagani, Chang Zhu, Lukas Dresel, G. Vigna, Christopher Kruegel, Ruoyu Wang, Tiffany Bao, Yan Shoshitaishvili, Adam Doupé","doi":"10.1109/SP46215.2023.10179317","DOIUrl":"https://doi.org/10.1109/SP46215.2023.10179317","url":null,"abstract":"Black-box web application vulnerability scanners attempt to automatically identify vulnerabilities in web applications without access to the source code. However, they do so by using a manually curated list of vulnerability-inducing inputs, which significantly reduces the ability of a black-box scanner to explore the web application’s input space and which can cause false negatives. In addition, black-box scanners must attempt to infer that a vulnerability was triggered, which causes false positives.To overcome these limitations, we propose Witcher, a novel web vulnerability discovery framework that is inspired by grey-box coverage-guided fuzzing. Witcher implements the concept of fault escalation to detect both SQL and command injection vulnerabilities. Additionally, Witcher captures coverage information and creates output-derived input guidance to focus the input generation and, therefore, to increase the state-space exploration of the web application. On a dataset of 18 web applications written in PHP, Python, Node.js, Java, Ruby, and C, 13 of which had known vulnerabilities, Witcher was able to find 23 of the 36 known vulnerabilities (64%), and additionally found 67 previously unknown vulnerabilities, 4 of which received CVE numbers. In our experiments, Witcher outperformed state of the art scanners both in terms of number of vulnerabilities found, but also in terms of coverage of web applications.","PeriodicalId":439989,"journal":{"name":"2023 IEEE Symposium on Security and Privacy (SP)","volume":"288 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122087215","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/SP46215.2023.10179310
Shubham Jain, Ana-Maria Creţu, Antoine Cully, Yves-Alexandre de Montjoye
End-to-end encryption (E2EE) provides strong technical protections to individuals from interferences. Governments and law enforcement agencies around the world have however raised concerns that E2EE also allows illegal content to be shared undetected. Client-side scanning (CSS), using perceptual hashing (PH) to detect known illegal content before it is shared, is seen as a promising solution to prevent the diffusion of illegal content while preserving encryption. While these proposals raise strong privacy concerns, proponents of the solutions have argued that the risk is limited as the technology has a limited scope: detecting known illegal content. In this paper, we show that modern perceptual hashing algorithms are actually fairly flexible pieces of technology and that this flexibility could be used by an adversary to add a secondary hidden feature to a client-side scanning system. More specifically, we show that an adversary providing the PH algorithm can "hide" a secondary purpose of face recognition of a target individual alongside its primary purpose of image copy detection. We first propose a procedure to train a dual-purpose deep perceptual hashing model by jointly optimizing for both the image copy detection and the targeted facial recognition task. Second, we extensively evaluate our dual-purpose model and show it to be able to reliably identify a target individual 67% of the time while not impacting its performance at detecting illegal content. We also show that our model is neither a general face detection nor a facial recognition model, allowing its secondary purpose to be hidden. Finally, we show that the secondary purpose can be enabled by adding a single illegal looking image to the database. Taken together, our results raise concerns that a deep perceptual hashing-based CSS system could turn billions of user devices into tools to locate targeted individuals.
{"title":"Deep perceptual hashing algorithms with hidden dual purpose: when client-side scanning does facial recognition","authors":"Shubham Jain, Ana-Maria Creţu, Antoine Cully, Yves-Alexandre de Montjoye","doi":"10.1109/SP46215.2023.10179310","DOIUrl":"https://doi.org/10.1109/SP46215.2023.10179310","url":null,"abstract":"End-to-end encryption (E2EE) provides strong technical protections to individuals from interferences. Governments and law enforcement agencies around the world have however raised concerns that E2EE also allows illegal content to be shared undetected. Client-side scanning (CSS), using perceptual hashing (PH) to detect known illegal content before it is shared, is seen as a promising solution to prevent the diffusion of illegal content while preserving encryption. While these proposals raise strong privacy concerns, proponents of the solutions have argued that the risk is limited as the technology has a limited scope: detecting known illegal content. In this paper, we show that modern perceptual hashing algorithms are actually fairly flexible pieces of technology and that this flexibility could be used by an adversary to add a secondary hidden feature to a client-side scanning system. More specifically, we show that an adversary providing the PH algorithm can \"hide\" a secondary purpose of face recognition of a target individual alongside its primary purpose of image copy detection. We first propose a procedure to train a dual-purpose deep perceptual hashing model by jointly optimizing for both the image copy detection and the targeted facial recognition task. Second, we extensively evaluate our dual-purpose model and show it to be able to reliably identify a target individual 67% of the time while not impacting its performance at detecting illegal content. We also show that our model is neither a general face detection nor a facial recognition model, allowing its secondary purpose to be hidden. Finally, we show that the secondary purpose can be enabled by adding a single illegal looking image to the database. Taken together, our results raise concerns that a deep perceptual hashing-based CSS system could turn billions of user devices into tools to locate targeted individuals.","PeriodicalId":439989,"journal":{"name":"2023 IEEE Symposium on Security and Privacy (SP)","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114762602","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：