do you have some statistical data about what percentage of the Earth's surface is covered by OSM data?

It depends - "covered" is an imprecise term. There's 100% coverage at country level, but that's not a too precise sort of coverage ;) But even in the case of well mapped regions, some places are only mapped as "there's a village here" - does that count as coverage? Does "major streets mapped" count as coverage? Is "building outlines mapped" sufficient, without an indication of their purpose? Even in inhabited regions, not all rivers are fully mapped - does that count against coverage, somehow? It probably depends a lot on the purpose of the coverage.

I am thinking about buildings, especialy in poor areas... in terms of doing something (automate) to put the world's vulnerable communities on the map. I also saw some statistical data about members, ways etc... and I was thinking If you have something about areas too.... - how many skm are mapped? - how much remained?

:)

What's really important here is "quality". Knowing that there's a building at a certain lat/long isn't especially helpful; but knowing that it's the only health centre for miles around is really useful.

I suspect you'd need to think about "percentage of facility type X in country Y that are actually mapped" and "accuracy with which they are mapped", both in terms of geographical accuracy, and also in terms of classification (is something a "doctors" or a "hospital", for example?).

If you define your question a bit more, it might be possible for someone to answer it.

Please note that we have strict rules about automated editing, importing, or adding of data. OSM's strength is human-added data. If you're thinking about some sort of algorithm that would detect buildings from imagery and add them automatically, that would most likely not be acceptable at all for automatic importing. There are however editor plugins already that auto-trace building outlines and therefore make it easier for human contributors to add data.

It is rather obvious where the "world's vulnerable communities" comes from. That is quite a worthy goal, even though I would prefer the "vulnerable" communities putting themselves on the map and not people with good intentions doing it for them.

However in any case it should be noted that adding lots of buildings is a relative recent phenomenon in OSM and it is quite possible to have a detailed, very helpful and useful map without a single building outline (for some reason adding lots of eye candy tends to be high on the priority list for humanitarian organisations). Even more so, adding lots of unnecessary, not really particularly important or permanent features make the map harder to maintain, particularly if the "vulnerable communities" are supposed to be the "owner" of the map in their area.