Join this courtroom-themed conference talk where Annie Talvasto from Waovo and Karl Ots from EPAM Systems put common AI security mistakes on trial. Explore vulnerabilities in AI deployments including exposed sensitive data, flawed model training, and insecure pipelines. Discover how Kubeflow, confidential computing, and other open source projects can defend your AI infrastructure by enforcing security guardrails, protecting sensitive data, ensuring compliance, and mitigating operational risks. This 26-minute CNCF presentation combines technical insights with engaging courtroom drama to help identify and address AI security challenges, enabling you to build secure, scalable AI pipelines with confidence.