End-to-End Secure MLOps: Leveraging Generative AI for Automated Threat Simulation and Continuous Security Validation
Year: 2024
Type: article
Abstract: The increasing adoption of Machine Learning Operations (MLOps) pipelines introduces both efficiency and new security risks to enterprise AI deployments. While MLOps accelerates model development and deployment, the complexity of continuous integration and delivery (CI/CD) in AI systems leaves potential attack surfaces, including data poisoning, model inversion, and supply chain vulnerabilities. This paper explores the integration of generative AI models for automated threat simulation within secure MLOps workflows, enabling continuous validation and proactive defense against adversarial threats. We propose an architecture that combines generative adversarial models for attack emulation with automated penetration testing for MLOps pipelines, ensuring end-to-end security. The work is positioned in the context of early MLOps adoption circa, with emphasis on preemptive threat discovery.
Author: Researcher
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W6949740055
FWCI:
Cites:
Cited by:
Related to: