An official website of the United States government
While Kubernetes enables practitioners to rapidly deploy their software and perform container orchestration efficiently, security of the Kubernetes-based deployment infrastructure is a concern for industry practitioners. A systematic understanding of how dynamic analysis can be used for securing Kubernetes deployments can aid practitioners in securing their Kubernetes deployments. We present an experience report, where we describe empirical findings from three dynamic application security testing (DAST) tools on a Kubernetes deployment used by 'Company-Z'. From our empirical study, we find (i) 3,442 recommended security configurations are violated in 'Company-Z's' Kubernetes deployment; and (ii) of the three studied DAST tools, Kubescape and Kubebench provide the highest support with respect to detecting 14 types of recommended security configurations. Based on our findings, we recommend practitioners to apply DAST tools for their Kubernetes deployments, and security researchers to investigate how to detect configuration violations dynamically in the Kubernetes deployment.
more »
« less
This content will become publicly available on May 3, 2026
An Exploratory Study of Security Vulnerabilities in Machine Learning Deployment Projects
Machine learning (ML) deployment projects are used by practitioners to automatically deploy ML models. While ML deployment projects aid practitioners, security vulnerabilities in these projects can make ML deployment infrastructure susceptible to security attacks. A systematic characterization of vulnerabilities can aid in identifying activities to secure ML deployment projects used by practitioners. We conduct an empirical study with 149 vulnerabilities mined from 12 open source ML deployment projects to characterize vulnerabilities in ML deployment projects. From our empirical study, we (i) find 68 of the 149 vulnerabilities are critically or highly severe; (ii) derive 10 consequences of vulnerabilities, e.g., unauthorized access to trigger ML deployments; and (iii) observe established quality assurance activities, such as code review to be used in the ML deployment projects. We conclude our paper by providing a set of recommendations for practitioners and researchers. Dataset used for our paper is available online.
more »
« less
- Award ID(s):
- 2312321
- PAR ID:
- 10635996
- Publisher / Repository:
- IEEE
- Date Published:
- ISBN:
- 979-8-3315-1468-6
- Page Range / eLocation ID:
- 33 to 36
- Format(s):
- Medium: X
- Location:
- Ottawa, ON, Canada
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Machine learning (ML) operations or MLOps advocates for integration of DevOps- related practices into the ML development and deployment process. Adoption of MLOps can be hampered due to a lack of knowledge related to how development tasks can be automated. A characterization of bot usage in ML projects can help practitioners on the types of tasks that can be automated with bots, and apply that knowledge into their ML development and deployment process. To that end, we conduct a preliminary empirical study with 135 issues reported mined from 3 libraries related to deep learning: Keras, PyTorch, and Tensorflow. From our empirical study we observe 9 categories of tasks that are automated with bots. We conclude our work-in-progress paper by providing a list of lessons that we learned from our empirical study.more » « less
-
null (Ed.)The ubiquitous usage of robots in modern society necessitates secure development of robotics systems. Practitioners who engage in robot development can benefit from a systematic study that investigates the categories of vulnerabilities that appear in robotics systems. The goal of this paper is to help practitioners mitigate vulnerabilities in robotics systems by conducting an empirical study of vulnerabilities in robotics systems. We conduct an empirical study where we analyze 176 robotics-related vulnerabilities collected from the Robot Vulnerability Database (RVD). Our findings show that: (i) robotics-related vulnerabilities can be classified into nine categories; (ii) memory-related vulnerabilities are the most frequent category, (iii) 92.6% of the reported vulnerabilities are software-related, and (iv) software components in robotics systems include more critical vulnerabilities compared to that of hardware components. Based on our findings, we provide a list of development activities that can be used to mitigate vulnerabilities for robotics systems.more » « less
-
Context: Supervised learning-based projects (SLPs), i.e., software projects that use supervised learning algorithms, such as decision trees are useful for performing classification-related tasks. Yet, security weaknesses, such as the use of hard-coded passwords in SLPs, can make SLPs susceptible to security attacks. A characterization of security weaknesses in SLPs can help practitioners understand the security weaknesses that are frequent in SLPs and adopt adequate mitigation strategies. Objective: The goal of this paper is to help practitioners securely develop supervised learning-based projects by conducting an empirical study of security weaknesses in supervised learning-based projects. Methodology: We conduct an empirical study by quantifying the frequency of security weaknesses in 278 open source SLPs. Results: We identify 22 types of security weaknesses that occur in SLPs. We observe ‘use of potentially dangerous function’ to be the most frequently occurring security weakness in SLPs. Of the identified 3,964 security weaknesses, 23.79% and 40.49% respectively, appear for source code files used to train and test models. We also observe evidence of co-location, e.g., instances of command injection co-locates with instances of potentially dangerous function. Conclusion: Based on our findings, we advocate for a shift left approach for SLP development with security-focused code reviews, and application of security static analysis.more » « less
-
Despite being beneficial for rapid delivery of software, Kubernetes deployments can be susceptible to security attacks, which can cause serious consequences. A systematic characterization of how community-prescribed security configurations, i.e., security configurations that are recommended by security experts, can aid practitioners to secure their Kubernetes deployments. To that end, we conduct an empirical study with 53 security configurations recommended by the Center for Internet Security (CIS), 20 survey respondents, and 544 configuration files obtained from the open source software (OSS) and proprietary domains. From our empirical study, we observe: (i) practitioners can be unaware of prescribed security configurations as 5% ~40% of the survey respondents are unfamiliar with 16 prescribed configurations; and (ii) for Company-A and OSS respectively, 18.0% and 17.9% of the configuration files include at least one violation of prescribed configurations. From our evaluation with 5 static application security testing (SAST) tools we find (i) only Kubescape to support all of the prescribed security configuration categories; (ii) the highest observed precision to be 0.41 and 0.43 respectively, for the Company-A and OSS datasets; and (iii) the highest observed recall to be respectively, 0.53 and 0.65 for the Company-A and OSS datasets. Our findings show a disconnect between what CIS experts recommend for Kubernetes-related configurations and what happens in practice. We conclude the paper by providing recommendations for practitioners and researchers. Dataset used for the paper is publicly available online.more » « less
