The emerging aircraft autonomy technologies powered by AI models are expected to change the aviation landscape. These AI models have been tested and deployed on commercial airliners, electric vertical takeoff and landing (eVTOL) aircraft, and small unmanned aerial systems (UAS). In this talk, I will briefly introduce two open-source AI models and their aviation use cases: (1) deep reinforcement learning for aircraft separation assurance and (2) deep neural networks for computer-vision-based autonomous landing. More importantly, I will cover several lines of work in AI safety and certification from my research lab, industry partners and government agencies in the U.S. aviation and avionics communities.