Your organization is creating a real-time recommendation system using machine learning models that analyze live user activity data stored in BigQuery and Cloud Storage. Each new model is stored in Artifact Registry, and the system deploys models to Google Kubernetes Engine, with Pub/Sub managing message queues. Recently, there have been reports of attacks targeting machine learning model supply chains. You need to improve the security of this serverless architecture, focusing on the development and deployment pipeline. What should you do to strengthen security?