Project Amplify aims to scale expert community-based feedback to improve large language models with goals of identifying potential harms in Africa related to LLMs and creating high-quality datasets in different African languages for tuning, reinforcement learning, and adversarial testing.