Yinzhi Cao

Yinzhi Cao

Auto-generated profile, based on publicly available information

Yinzhi Cao

Johns Hopkins University

Research Focus & Metrics

Academic Career Length

13Years (2014-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
2 December 2024
Verified
Conference Paper
Open access
BHBo Hui
Bo Hui
HYHaolin Yuan
Haolin Yuan
NGNeil Gong
Neil Gong
PBPhilippe Burlina
Philippe Burlina
YCYinzhi Cao
Yinzhi Cao
Large Language Models (LLMs) enable a new ecosystem with many downstream applications, called LLM applications, with different natural language processing tasks. The functionality and performance of an LLM application highly depend on its system prompt, which instructs the backend LLM on what task to perform. Therefore, an LLM application developer often keeps a system prompt confidential to protect its intellectual property. As a result, a natural attack, called prompt leaking, is to steal the system prompt from an LLM application, which compromises the developer's intellectual property. Existing prompt leaking attacks primarily rely on manually crafted queries, and thus achieve limited effectiveness. In this paper, we design a novel, closed-box prompt leaking attack framework, called PLeak, to optimize an adversarial query such that when the attacker sends it to a target LLM application, its response reveals its own system prompt. We formulate finding such an adversarial query as an optimization problem and solve it with a gradient-based method approximately. Our key idea is to break down the optimization goal by optimizing adversary queries for system prompts incrementally, i.e., starting from the first few tokens of each system prompt step by step until the entire length of the system prompt. We evaluate PLeak in both offline settings and for real-world LLM applications, e.g., those on Poe, a popular platform hosting such applications. Our results show that PLeak can effectively leak system prompts and significantly outperforms not only baselines that manually curate queries but also baselines with optimized queries that are modified and adapted from existing jailbreaking attacks. We responsibly reported the issues to Poe and are still waiting for their response. Our implementation is available at this repository: https://github.com/BHui97/PLeak.

1 November 2024
Verified
Book Chapter
Closed access
PDF file
YYYuchen Yang
Yuchen Yang
KLKwonjoon Lee
Kwonjoon Lee
BDBehzad Dariush
Behzad Dariush
YCYinzhi Cao
Yinzhi Cao
SLShao-Yuan Lo
Shao-Yuan Lo

7 January 2022 in IEEE Transactions on Dependable and Secure Computing
Verified
Journal Article
Closed access
LJLing Jin
Ling Jin
YCYinzhi Cao
Yinzhi Cao
YCYan Chen
Yan Chen
DZDi Zhang
Di Zhang
SCSimone Campanoni
Simone Campanoni
Smart contracts, just like other computer programs, are prone to a variety of vulnerabilities, which lead to severe consequences including massive token and coin losses. Prior works have explored automated exploit generation for vulnerable Ethereum contracts. However, the scopes of prior works are limited in both vulnerability types and contract platforms. In this paper, we propose a cross-platform framework, called EXGEN, to generate multiple transactions as exploits to given vulnerable smart contracts. EXGEN first translates either Ethereum or EOS contracts to an intermediate representation (IR). Then, EXGEN generates symbolic attack contracts with transactions in a partial order and then symbolically executes the attack contracts together with the target to find and solve all the constraints. Lastly, EXGEN concretizes all the symbols, generates attack contracts with multiple transactions, and verifies the generated contracts' exploitability on a private chain with values crawled from the public chain. We implemented a prototype of EXGEN and evaluated it on Ethereum and EOS benchmarks. EXGEN successfully exploits 1,258/1,399 (89.9%) Ethereum and 126/130 (96.9%) EOS vulnerabilities. EXGEN is also able to find zero-day vulnerabilities on EOS.

25 January 2022 in IEEE Transactions on Dependable and Secure Computing
Verified
Journal Article
Closed access
XZXiaohan Zhang
Xiaohan Zhang
MZMi Zhang
Mi Zhang
YZYuan Zhang
Yuan Zhang
MZMing Zhong
Ming Zhong
XZXin Zhang
Xin Zhang
MYMin Yang
Min Yang
Learning-based malware detectors are widely used in practice to safeguard real-world computers. One major challenge is known as model aging, where the effectiveness of these models drops drastically as malware variants keep evolving. To tackle model aging, most existing works choose to label new samples to retrain the aged models. However, such data-perspective methods often require excessive costs in labeling and retraining. In this paper, we observe that during evolution, malware samples often preserve similar malicious semantics while switching to new implementations with semantically equivalent APIs. Such observation enables us to look into the problem from a different perspective: feature space. More specifically, if the models can capture the intrinsic semantics of malware variants from feature space, it will help slow down the aging of learning-based detectors. Based on this insight, we design APIGRAPH to automatically extract API knowledge from API documentation and incorporate these knowledge into the training of malware detection models. We use APIGRAPHto enhance 5 state-of-the-art malware detectors, covering both Android and Windows platforms and various learning algorithms. Experiments on large-scale, evolutionary datasets with nearly 340K samples show that APIGRAPH can help slow down the aging of these models by 5.9% to 19.6%, as well as reduce labeling efforts from 33.07% to 96.30%on top of data-perspective methods.

8 December 2014
Verified
Conference Paper
Open access
YCYinzhi Cao
Yinzhi Cao
XPXiang Pan
Xiang Pan
Yan ChenYan Chen
Yan Chen
JZJianwei Zhuge
Jianwei Zhuge
Drive-by download attacks, which exploit vulnerabilities of web browsers to control client computers, have become a major venue for attackers. To detect such attacks, researchers have proposed many approaches such as anomaly-based [22, 23] and vulnerability-based [44, 50] detections. However, anomaly-based approaches are vulnerable to data pollution, and existing vulnerability-based approaches cannot accurately describe the vulnerability condition of all the drive-by download attacks. In this paper, we propose a vulnerability-based approach, namely JShield, which uses novel opcode vulnerability signature, a deterministic finite automaton (DFA) with a variable pool at opcode level, to match drive-by download vulnerabilities. We investigate all the JavaScript engine vulnerabilities of web browsers from 2009 to 2014, as well as those of portable document files (PDF) readers from 2007 to 2014. JShield is able to match all of those vulnerabilities; furthermore, the overall evaluation shows that JShield is so lightweight that it only adds 2.39 percent of overhead to original execution as the median among top 500 Alexa web sites.

15 November 2023
Verified
Conference Paper
Open access
JYJianjia Yu
Jianjia Yu
SLSong Li
Song Li
JZJunmin Zhu
Junmin Zhu
YCYinzhi Cao
Yinzhi Cao
Extensions complement web browsers with additional functionalities and also bring new vulnerability venues, allowing privilege escalations from adversarial web pages to use extension APIs. Prior works on extension vulnerability detection adopt classic static analysis, which is unable to handle dynamic JavaScript features such as those function calls as part of array lookups. At the same time, prior abstract interpretation focuses on lightweight server-side JavaScript, which often cannot scale to client-side extension code due to object explosions in the abstract domain. In this paper, we design, implement and evaluate a novel, coverage-driven, concurrent abstract interpretation framework, called CoCo, to efficiently detect vulnerabilities in browser extensions. On one hand, CoCo parallelizes abstract interpretation with concurrent taint propagation for each branching statement, message passing and content/background scripts to detect vulnerabilities with improved scalability. On the other hand, CoCo prioritizes analysis that increases code coverage, thus further detecting more vulnerabilities. Our evaluation shows that CoCo detects at least 43 zero-day, exploitable, manually-verified extension vulnerabilities that cannot be detected by state-of-the-art works. We responsibly disclosed all the zero-day vulnerabilities to extension developers.