A Generative AI Engineer is preparing to deploy an LLM application built with Foundation Model APIs and wants to adhere to security best practices for a production environment. Which authentication method should they select?