Are you struggling to keep up with the fast-paced world of AI security? Join us as Microsoft MVP Martin Kimbowski breaks down how to protect your machine learning environments and master DevSecOps in the Microsoft cloud. This episode is a deep dive into the intersection of artificial intelligence and cybersecurity.
In this conversation, Martin shares insights from his 20-year career in the Microsoft ecosystem. We explore why organizations must move beyond simple development and embrace a secure by design philosophy. You will learn about the mechanics of prompt injection, the dangers of API key leakage, and how to use threat modeling to identify gaps before they become critical failures.
Martin also explains how to leverage the Microsoft security stack, including Defender for Cloud, Entra ID, and GitHub Advanced Security, to create robust AI pipelines. Whether you are a developer, a data scientist, or a security professional, this episode provides a roadmap for balancing innovation with protection. We wrap up with practical tips for career growth and why continuous learning is the ultimate defense in the age of AI.
Chapters
0:00 Introduction to Martin Kimbowski
2:45 The shift from DevOps to AI security
6:15 Explaining MLOps and Secure by Design
9:30 Biggest security mistakes in AI today
13:45 What is prompt injection?
18:15 Data leakage risks in AI pipelines
22:30 AI attack chains and malicious models
26:45 Threat modeling for machine learning
31:15 Security as a team sport vs bottleneck
35:30 Governance and Microsoft Purview
39:45 Protecting workloads with Microsoft Defender
43:15 GitHub Advanced Security and Entra ID
48:00 Best practices for access management
51:45 Rapid fire round and final thoughts
If you found this helpful, please like the video and subscribe for more cloud security content.
#AISecurity #MicrosoftCloud #DevSecOps #MLOps #CyberSecurity
