ABSTRACT We investigate the population of bright satellites ($M_{*} \ge 10^{5} \, \mathrm{M}_{\odot }$) of haloes of mass comparable to that of the Milky Way in cosmological simulations in which the dark matter (DM) is either cold, warm, or self-interacting (CDM, WDM, and SIDM, respectively). The nature of the DM gives rise to differences in the abundance and structural properties of field haloes. In WDM, the main feature is a reduction in the total number of galaxies that form, reflecting a suppression of low-mass DM haloes and lower galaxy formation efficiency compared to CDM. For SIDM, the changes are structural, restricted to the central regions of haloes and dependent on the assumed self-interaction cross-section. We also consider different baryonic subgrid physics models for galaxy formation, in which supernova gas blowouts can or cannot induce the formation of a core in dwarf galaxies. Overall, the inclusion of baryons lessen the differences in the halo properties in the different DM models compared to DM-only simulations. This affects the satellite properties at infall and therefore their subsequent tidal stripping and survival rates. None the less, we find slightly less concentrated satellite radial distributions as the SIDM cross-section increases. Unfortunately, we also find that the satellite populations in simulations with baryon-induced cores in CDM and WDM can mimic the results found in SIDM, making the satellite stellar mass and maximum circular velocity functions heavily degenerate on the assumed nature of the DM and the adopted subgrid modelling. These degeneracies preclude using the brightest satellites of the Milky Way to constrain the nature of DM.