Explore the critical importance of MLSecOps in AI Security Posture Management through this insightful 43-minute podcast episode featuring Sean Morgan, Chief Architect at Protect AI. Delve into realistic AIML security risks, learn how to measure and improve your organization's AI Security Posture, and understand the integration of security practices into the AIML lifecycle. Gain valuable insights on topics such as malicious data concerns during training, model artifact security, vulnerabilities in inference servers and popular ML tools, and the role of automated scanning in enhancing protection. Discover how balancing ML and security processes can lead to robust AI systems while addressing the nuances of model building, security, and working with foundational models.
MLSecOps: Fundamental to Robust AI Security Posture Management - Podcast #257
Overview
Syllabus
[] Sean's preferred coffee
[] Takeaways
[] Register for the Data Engineering for AI/ML Conference now!
[] KubeCon Paris: Emphasis on security and AI
[] Concern about malicious data during training process
[] Model builders, security, pulling foundational models, nuances
[] Hugging Face research on security issues
[] Inference servers exposed; potential for attack
[] Balancing ML and security processes for ease
[] Model artifact security in enterprise machine learning
[] Scanning models and datasets for vulnerabilities
[] Ray's user interface vulnerabilities lead to attacks
[] ML Flow vulnerabilities present significant server risks
[] Data ops essential for machine learning security
[] Prioritized security in model and data deployment
[] Automated scanning tool for improved antivirus protection
[] Wrap up
Taught by
MLOps.community
