Artificial Intelligence has come into focus anew in the context of digitization and global competition. So has the tension between human ethics, regulation, and the potential gains this technology field offers for economic and societal progress. This paper is intended to contribute to the ongoing debate about opportunities and uncertainties in particular with respect to the use of private data in AI. We discuss the status of AI outcomes in terms of their validity, and of AI input as to the quality of data. In a first order approach we distinguish between the commercial, public, industrial, and scientific data spheres of AI systems. We resume the ethical and regulative approaches to the utilization and protection of massive private data for AI. Regarding the currently favoured ways of organizing the collection and protection of data we refer to respective ruling and denominate distributed ledger systems and open data spaces as functional means. We conclude by arguing that governing data privacy and quality by distinguishing different AI data spheres will enable a reasonable balance of these two aspects.