Cloud Security Challenges in the AI Era

Marina Moore, a seasoned security researcher and CNCF security & compliance TAG co-chair, digs into why containers aren’t the silver bullet they’re often made out to be. She traces the root causes of container vulnerability—from shared kernels to weak runtime isolation—and argues that simply bolting on more scanning tools doesn’t cut it. As a more robust path forward, Marina suggests exploring micro-VMs to isolate workloads, tightening supply-chain checks, and embracing stronger policy enforcement.

On top of container risks, Marina flags AI inference as a growing attack vector. Models in production can inadvertently leak data or be poisoned by malicious inputs, widening your threat surface. Her takeaway? Rethink the “build and run anywhere” mantra with a lens on zero-trust isolation and continuous monitoring, especially as AI workloads surge.

Watch on YouTube