metallic covers of devices like MacBook Air should shield the interior from electromagnetic waves rather effectively.
Probably not as much as you'd think. If the covers surrounded the antenna on all sides and had no large (relative to wavelength) holes in them, then indeed they would be Faraday cages and very little (none, ideally) radiation would get in or out.
But the MacBook's case is not a Faraday cage. It has slots (for the CD), holes (for the keys, screen, cables), seams, and so on.
Indeed, the electric field radiated by the antenna is intercepted by the case. This electric field causes RF currents in the case, because the charge carriers (electrons) in the metal want to find the lowest possible electric potential. If the case has no holes and is very conductive, then they are able to rearrange completely such that the electric field is cancelled.
But if there are holes, the RF currents can't go through it. They have to go around, and this results in a less than complete cancellation of the electric field. The charges keep moving around the hole as the antenna's field changes, and you end up with moving charges (current) held apart by the hole (voltage), just as there would be in an antenna. Consequently, some of the energy is re-radiated.
In fact, if the hole is the right size, it can be just as efficient as an antenna. It's called a slot antenna. For some applications, RF engineers will intentionally create them because they are more convenient to fabricate than some other more familiar antenna, like a dipole. Slot antennas also have consequences for PCB designers, who must avoid unintentionally creating slot antennas (by introducing breaks into the ground plane, usually) which would cause their device to fail EMI requirements.
So there you have it. The case doesn't always block RF radiation. As Some Hardware Guy says, the product engineers just need to find a clever place to stick the antenna and ensure it's properly tuned.