The large-scale digitization of academic publications and patrimonial resources has recently stirred significant interests in text and data mining across scientific communities. Theses computing and statistical techniques allow to extract structured information or, more relevantly for the humanities, to map the textual characteristics (such as genres or intertextual relations) of very large corpora. Yet, the promises of text and data mining are constrained by legal uncertainties, mostly regarding intellectual property rights. Without copyright holders’ agreement, the automated recopy of lawful sources and their treatment by the members of a scientific project are very likely illegal. This study aims to address the main legal stakes of mining projects in social science and the humanities and to recount the gradual transformation of informal claims into structured mobilization to implement exceptions. In several countries such as the United States, Japan or the Canada, the right to mine has become a de facto extension of the right to read thanks to pre-existing exceptions. In the European Union, this process requires explicit legal reforms, as the legal frame of the 2001 author rights directive proves too restrictive. Since 2014, three major European countries have passed a text mining exception, the United Kingdom, Germany and France, with the French version remaining for the moment a partly failed attempt. In parallel with theses national initiatives, an European-wide exception will likely be part of the currently debated European Authors Rights Reform. Theses legal evolutions not only helped to secure text and data mining activities in research but seems to have encourage the structuration of emerging scientific practices, as the enforcement of the exception requires to codification common norms and infrastructures.