"Mapping data" is too broad of a term, because it probably includes a lot of copyrightable information. However, the factual basics, like GPS coords, are not copyrightable.
An individual data point maybe not, but even something like "there is a road between X1,Y1 and X2,Y2" you can't just extract from Google Maps and put in OSM, at least not multiple times. There are surprisingly few generally usable data sources (some government datasets that have been put under open licenses, Microsofts permits to use Bing's aerial photography as a basis for OSM, ...)
I'm not sure if OSM really needs to be as cautious as they are. I don't see how you could argue that "Main St begins at X1, Y1" and "Main St end at X2,Y2" is a single copyrightable work instead of 2 separate non-copyrightable facts. The sentence or package which contains these two separate data points may be a unique work that qualifies for copyright protection, but the raw data points aren't once you separate them out.
Open-source projects have a history of being hyper-sensitive to these concerns so that they never have to waste time with lawyers lobbing IP claims. A certain distribution derived from the sources released by a prominent North American enterprise Linux vendor comes to mind, as does WINE's refusal to accept any code from anyone who has any relationship with or connection to Microsoft whatsoever.
Google has clear TOS that disallow copying data so it isn't just about copyright, and EU database rules are anyway a much bigger concern than US copyright law.
OpenStreetMap needs to be useful for everyone, not just people in jurisdictions with more liberal copyright laws.
But you can't really separate these datapoints, at least not in a legally safe manner. If you take a "work", reduce it down to the basic data it probably was created from and rebuild a basically identical work, how do you defend against the claim that you copied it? You could try to go through multiple steps, done by different entities, and hope to hide that way, but coordinating that makes you vulnerable again.
How could a project like OSM make sure that it didn't happen on a large, clearly violating scale (because many people make copies of small parts, recreating the larger, protected work), especially if it isn't established what an internationally safe standard is.
In practice, you are right, you can copy a single street from google maps to OSM. I bet some mappers in the history of OSM have traced screenshots of google maps or something. But that is only safe because it can't be proven, unless you are unlucky enough to copy a canary.
If all you have is a compilation of facts, then yes, people can copy the data points one by one, alter the format slightly, and have a wholly separate work that has no legally recognized parentage. This is why you don't see a lot of plain collections of facts for sale.
As far as I know, Google Maps is protected from scripts copying the data points out only by its Terms of Use, which the CFAA basically eval()s into law.
Google Maps won't disappear overnight because they do provide a substantial amount of beneficial proprietary data, like the illustrations and private aerial/satellite imagery (imagery from sources like NASA is public ___domain), their scripts that allow easy embedding, the ability to connect to one's Google account, and so forth. But there is no legal protection of the raw data points used to build Google Maps as far as I know.
Let me reiterate that I'm not a lawyer and no one should do anything based on my posts.
