Towards Guaranteed Safe Ai: A Framework To Ensure Robust And Reliable Ai Systems