In recent years, FAIR principles have come a long way to serve the global need for generic guidelines governing data management and stewardship. Considering their wide embrace and the support received from governments, policy-makers, governing bodies and funding bodies, FAIR principles have all the potential to have a huge impact on the future landscape of knowledge creation for the better. This opportunity, however, may easily be missed if the specific dynamics of scientific production are not addressed in its disciplinary implementation plans. With the goal of making FAIR meaningful and helping to realise its promises in an arts and humanities context, this paper describes some of the defining aspects underlying the domain-specific epistemic processes that pose hidden or visible challenges in the FAIRification of knowledge creation in Arts and Humanities. By applying the FAIR data guiding principles to arts and humanities data curation workflows, we will show that contrary to their general scope and deliberately domain-independent nature, they have been implicitly designed along underlying assumptions about how knowledge creation operates and communicates. These are: 1. scholarly data or metadata is digital by nature, 2. scholarly data is always created and therefore owned by researchers, and 3. there is a wide community-level agreement on what can be considered scholarly data. The problems around such assumptions in arts and humanities are cornerstones in reconciling disciplinary traditions with the productive implementation of FAIR data management. By addressing them one by one, we aim to contribute to the better understanding of discipline-specific needs and challenges in data production, discovery and reuse. Based on these considerations, we make recommendations that may facilitate the inclusive and optimal implementation of the high-level principles that serve the flourishing of the arts and humanities disciplines rather than imposing limitations on its epistemic practices.