Cloud Security Challenges in the AI Era

Security researcher Marina Moore (co-chair of CNCF’s security and compliance TAG) warns that containers, despite their popularity, can inherit vulnerabilities from the host kernel and make it easier for attackers to break out, escalate privileges, or sneak in malicious code. She digs into common misconfigurations, offers hardening tips, and even suggests swapping out containers for micro-VMs when you need iron-clad isolation.

On the AI front, Moore points out that running inference at scale introduces fresh attack surfaces—everything from data poisoning to model theft—and urges architects to build in safeguards (like strict access controls and runtime checks) before you deploy your next AI service.

Watch on YouTube