Data extraction has long been a contested subject in big data environments. Advances in the production of synthetic data have given rise to new discussions around data ethics and politics. Synthetic data promises to bypass controversial aspects in data use, such as data anonymization and breaches as well as complement fairness, filling gaps in data sets for the sake of representation. However, there is an urgent need to question the sociopolitical implications of this emerging distinction between organic and synthetic data from critical data studies. This research talk presents critical considerations around the distinction between the real and “fake” in data structures. Drawing on early fieldwork, it provides reflections on the epistemologies of synthetic data, raising questions such as: In what way is the distinction between organic and synthetic the basis for the promising outcome of generated data? How do forms of generative AI potentially perpetuate structures of dominance?