The incorporation of artificial intelligence and machine learning into pharmaceutical formulation development has seen an increase in the last decade, enabling the promise of rapid, predictive, and less wasteful development of new medicines. In this paper, the authors perform a global comparative study of regulatory and ethical issues related to this change. The regulatory framework in the United States, European Union, Japan, China, and India is compared to determine where there is a consensus among countries' regulators about how they adapt their requirements of quality, validation, and post-market surveillance to the use of adaptive models based on data. Moreover, the paper considers general ethical issues of such a transition, including data privacy and origins, potential biases in algorithms and their effects on health equity, the conflict between transparency of models and their commercial interests, and assigning liability when the decisions of algorithms are a part of the reason behind a formulation failure or injury to a patient. It is concluded that international harmonisation, the governance of models longitudinally, and the consideration of ethics in existing quality-by-design practices are essential prerequisites for the global acceptance of artificial intelligence in formulating pharmaceuticals.