I don't work for a game studio, but I don't think 3d scanning is used very often in game development. The problem with it is simply that you get a very highly dense mesh. You then have to retopo this mesh. Also, the technology comes at a high price(I don't know numbers though). It would be financially better to model your things directly, using sketches for reference, though physical objects could also be used for reference as well, including sculpted clay. An exception to the above paragraph could be a specific style of art, which involves using clay models to create 2d sprites from pictures of said 3d objects. This has similar drawbacks to using 3d renders for 2d sprites, like lack of detail in lower resolution(without massive touchup).

Now, I'm sure there is at least a single studio somewhere for at the least a single game that has used this method. I couldn't confirm who because I don't know, but you would think that SOMEONE would try it right? Also, using real objects to aid in 3d design is found more often. Just as an example, animation can be done using sensors on an actual person, which would be connected to a rig in modeling software to create 3d animations, which would be much smoother and "realistic" than what an animator could generally create, though this depends on the skill of said animator.

It's extremely common for characters, even on medium-budget last-gen games. Even when I was working on PS2, we had laser scanned character face data.

Note though, that this is just a starting point for a 3D artist. The scanner data is never plugged straight into the game, it's just a shortcut to make an artist's job easier.

These days photogrammetry gets just as good results as laser scanning, but can deliver you texture data at the same time, so it's taking off in popularity (LA Noire's insanely detailed facial animations was based on 4D photogrammetry).

Also, generally you would hire some actors and rent time at another (probably film-VFX-oriented) studio who has a laser scanner... but with photogrammetry, game studios could feasibly build their own capture systems for a similar price.

Check out this article on the use of rented photo-scanning for Cyberpunk 2077: http://www.artofvfx.com/?p=3258

For environments and general props, this kind of scanning is not very common. If it is used, it would probably be low-quality photogrammetry conducted by your own staff.

As for the fact that you get insanely detailed models from these processes (which you need to retopo) - that's a really good thing.

Yes, this means that you basically have to remodel the object yourself with a low polygon count... but once you've done that, you can transfer over all the high-poly detail by baking it into a normal map, AO map, cavity map, etc...

Even when you're not using scanned data, this is a very common workflow -- artists sculpt an insanely high poly-count model in Z-brush etc (e.g. a million polygons), then create a low polygon version with a sensible topology (e.g. 10k polygons) and bake all the details onto this low-poly version.

As for animation based on sensors attached to a person (or from other camera systems, e.g. Kinect) - that's called mo-cap and it's also standard practice. Mo-cap is generally given to the animator as a starting point, and they use that initial data to create the game-ready animations.