The wording for change 1 looks fine to me.

For change 2, the mention of the internal representation is misleading, since the internal representation of (long) integers in current CPython is effectively sign-magnitude, and so there are some shenanigans to make the operations behave *as though* the internal representation were some form of two's complement [1]. The previously proposed wording (in msg321679) (with the "infinite sign bits" extracted into the main text as Tim suggests) looks fine to me.

[1] https://github.com/python/cpython/blob/c9265c1534b7e62bb9b15460d0420c0c3bb57ff9/Objects/longobject.c#L4415-L4429