There already exists at least one method that attempts to use electromagnetic precursors to earthquakes; it actually tries to detect electrical currents running up through active faults, explained by the piezoelectric phenomenon. Search for Varotsos if you want, perhaps there are other approaches as well.
Even though the device has a very long "antenna" (ie it tries to avoid noise which usually has shorter wavelengths) it has had some false positives (power lines may produce signals as well, they are long and signals may be of low frequency.
Personally I would never focus on short wavelengths such as TV frequencies. Varotsos correctly focused on low frequencies (there is a practical limit to the size of the "antenna", I think it is some 50-100 meters long). So in short, no, TV disturbances will never become a valid earthquake precursor. Too many false positives would ruin it.
Keep in mind that in the aforementioned approach, the first signal comes perhaps 4 months before the earthquake and further signals and seismic data are used in order to further narrow the time window, which right now is in the order of days to weeks. However the location and depth are pretty inaccurate (100 km radius or more) while the magnitude may be almost 1 unit lower (the most recent "success" had a magnitude 0.8 lower but it was a doublet). Currently there is an active "prediction" for an earthquake 100 km around Athens, of magnitude near 6. If it is to come true, it should happen until June or early July, but there are dozens of capable faults within the radius so it is practically useless at this point.