Two things. Firstly, although the anti-M may be "naturally occurring", it need not necessarily be IgM. Many examples of anti-M, reacting in the cold and RT are, in fact, agglutinating IgG (a bit like ABO IgG antibodies and anti-P in the DL Test). In addition, maternal IgG antibodies are actively transported across the placenta and, as a result, the baby ends up with a higher concentration of some IgG specificities in their circulation than is in the mother at birth. As the anti-M in the baby's circulation is so weak (showing dosage), it could be that it is of maternal origin, but that it is so weak now in the mother that it is undetectable in her plasma using normal serological techniques.
Secondly, some precocious babies are (rarely) known to produce their own IgM antibodies at birth (this can be proved by the babies producing an ABO antibody that cannot have come from the mother - say a group O baby with anti-B in the plasma, from a group B mother, or by looking at the Gm and/or Km types of the immunoglobulins - see, for example, Toivanen P, Hirvonen T. Iso- and heteroagglutinins in human fetal and neonatal sera. Scand J Haemat. 1969; 6: 42-48), so it could be the baby's own antibody, despite what I said above!
Lastly, and this came to me as I was typing (hence a third point), is the mother of the baby of Japanese origin? The only reason I ask is that, if she is, be a bit careful, as anti-M of low titre has been known to cause a sort of delayed HDFN in this ethnicity (see Yasuda H, Ohto H, Nollet KE, Kawabata K, Saito S, Yagi Y, Neggishi Y, Ishida A. Hemolytic disease of the fetus and newborn with late-onset anemia due to anti-M: a case report and review of the Japanese literature. Transfusion Medicine Reviews 2014; 28: 1-6. doi: 10.1016/j.tmrv.2013.10.002).
Incidentally, although I have seen anti-A or anti-B in a newborn's plasma, I have never seen an anti-M.