There have been observations, first from the MAGIC Telescope (July 2005) and quite recently (September 2008) from the FERMI Satellite Telescope, on non-simultaneous arrival of high-energy photons from distant celestial sources. In each case, the highest energy photons were delayed, as compared to their lower-energy counterparts, by clearly observable time intervals. Although the astrophysics at the source of these energetic photons is still not understood, and such non simultaneous arrival might be due to non simultaneous emission as a result of conventional physics effects, nevertheless, rather surprisingly, the observed time delays can also fit excellently some scenarios in quantum gravity, predicting Lorentz violating space-time "foam" backgrounds with a non-trivial subluminal vacuum refractive index suppressed linearly by a quantum gravity scale of order of the reduced Planck mass (∼ 1018 GeV). In this talk, I discuss the MAGIC and FERMI findings in this context. First, I review the high-energy astrophysics models for cosmic acceleration at celestial sources, stressing that currently there is no consensus as regards the observed delays. Then I derive estimates/bounds on the quantum gravity scale that reproduces the observed time delays on the assumption of a vacuum refractive index for photons with linear suppression, and argue on the consistency of such bounds with measurements from other Gamma Ray Telescopes, such as H.E.S.S. I then explain under which circumstances the MAGIC and FERMI findings could be accommodated in such models in agreement with all the other, currently available, astrophysics constraints of Lorentz Violation. The key features are: (i) transparency of the foam to electrons, (ii) absence of birefringence effects and (iii) a breakdown of the local effective lagrangian formalism. In contrast to other Quantum Gravity (field-theoretic) models with non-trivial optical properties available to date, a string model based on brane-worlds with the bulk space being punctured by space-time point-like D0-brane defects, that provide the seeds for Lorentz violating foamy structures, seems to respect all three requirements. The model provides an explanation for the observed photon time delays in a natural range of the string coupling and mass scale.