Can the "doublet-triplet splitting" problem be solved without doublet-triplet splitting?

We consider a new possible mechanism for the natural solution of the doublet-triplet splitting problem in SUSY GUTs. In contrast to the usually discussed scenarios, in our case the GUT symmetry breaking does not provide any splitting between the Higgs doublet and the triplet masses. The weak doublet and its colour triplet partner both remain light, but the triplet automatically occurs decoupled from the quark and lepton superfields and cannot induce proton decay. The advantage of the above scenario is the absence at the GUT scale of the baryon number violating the tree level d = 5 and d = 6 operators via the colour triplet exchange. It is shown that in flipped SU(5) GUT they do not appear at any scale. In the SO(10) model, such operators can be induced after SUSY breaking but are strongly suppressed.