As AI becomes integral to enterprise applications, securing AI/ML systems is paramount. While MITRE ATLAS provides a robust framework for understanding adversarial threats to AI, enterprises often struggle to adopt it in cloud-native
environments. This session demonstrates practical strategies for using open-source tools to operationalize AI security and align with the MITRE ATLAS framework.
We'll explore real-world use cases illustrating how tools like Kubeflow, Clair, Falco, StackRox, and Kubescape can be combined to detect and mitigate threats such as data poisoning, model extraction, and evasion attacks throughout the AI
lifecycle—from training to inference.
Attendees will gain insights into:
Understanding the MITRE ATLAS framework and its significance for AI/ML security, Mapping open-source tools to the ATLAS matrix for actionable, layered defenses, Integrating security controls into MLOps pipelines using Kubernetes-native tooling.