Advanced LIGO and Advanced Virgo detected dozens of gravitational wave (GW) signals from colliding black holes and neutron stars during its third observation run. Accurate sky localization and the rapid release of GW candidates are crucial for successful follow-up observations of possible electromagnetic (EM) emissions from these sources. While more accurate sky localization can be achieved with a global network of GW detectors that is increasing in size, there remain significant challenges facing the latter. I will discuss some of the challenges of correctly identifying gravitational wave signals from a population of non-Gaussian noise (glitches) in low-latency. I will also talk about the prospects of using the machine learning classifier, GWSkyNet, for overcoming these challenges and facilitating EM follow-up observations of GW sources that may be detected in the next upcoming observation run of the LIGO-VIRGO-KARGA collaboration.