I'm thinking of getting started with using Magic Lantern on the 6D. Troubling part is that ML mentions on its website that the software may brick the camera.
You have to differentiate between "crash" and "brick". Here's my experience with ML after 2.5 years on 60d 3 month 6d:
* "crash" means that the camera freezes. This can happen esp. in early builds due to bugs in ML and is not something to be overly concerned about because you just have to take out the battery (or just open the battery door), re-insert it, done. This lack of stability means that you shouldn't use pre-final ML in production environments where reliability is paramount, for example shooting a wedding.
* "brick" means your camera doesn't react or start anymore, even with a card w/o ML on it. Since the general ML code base is rather mature by now and most of the devs are very concerned about stability, it is very unlikely that you will ever experience this even running the latest nightly ML - I would dare to say that actual hardware damage for example by too high digic/cmos temperature is nearly impossible by now.
The main reason why bricking happens is that ML writes values to the Canon nvram that are out of bounds, and in most cases you can un-brick your camera with a special debug ML that you just have to run to clear the Canon nvram. A 100% foolproof way is to use a C mode since no settings are written to the Canon nvram, that's what I do when I code for ML and try out new changes.
Bottom line: Get it if you want to experiment a bit, aren't concerned about the lack of documentation on the latest features but want to get a great enhancement or script/code what your camera does yourself (and I'm only (still) using Canon because of this). If you are rather conservative about features and automatisation, don't feel limited by the Canon fw or are very fearful about your €3000 camera, don't use it.
Last not least: Labels like "Alpha" mostly mean not dangerous, but that it's not feature-complete. The dev resources of ML are very limited (it's essentially 2 people who really know how to do it) and the number of supported cameras grows, so don't expect a final 5d3 or 6d in the near future, not to speak of a 70d version.