AI-driven software delivery introduces new risks, especially prompt manipulation within CI/CD workflows. This article details the emerging fields of PromptOps and MLSecOps and offers practical strategies for securing prompts, models, and pipelines.