Good question. The semantics of the word "recent", in general, and in academic writing, in particular, is not clearly defined (that is, fuzzy), which makes its practical use quite tricky, as evidenced by your question.
While @vonbrand's answer offers some valuable insights, such as considering the fluidity of a particular scientific field or domain, I would suggest a more practical solution to this problem, as follows. Consider literature that you reference in a particular paper. What is the temporal range of the sources? I think that this aspect could guide you in to where the word "recent" is appropriate and where not so much.
For example, if you cite sources from the current century as well as 1930s, then a paper from 2010 should be considered recent, but not one from 1950. If, on the other hand, your temporal range of references is rather narrow, say, recent 20 years, then you should refer to as "recent" for sources that are from approximately last 4-5 years. You can come up with your own rule of thumb (10-20% of the total range sounds pretty reasonable). The most important aspect would be not the actual value (for the rule of thumb), but rather your consistency in applying it throughout the paper.