The type parameters ofnp.number and its abstract subtypes now default toAny.
This means that typecheckers will now infernp.floating asnp.floating[Any], instead of rejecting it (when configured in strict mode).

This affects

• np.number
• np.integer
• np.signedinteger
• np.unsignedinteger
• np.inexact
• np.floating
• np.complexfloating (by building uponTYP: Optional 2ndnumpy.complexfloating type parameter #27420)

The reason for choosingAny as the default, instead of setting it to the upper bound (numpy.typing.NBitBase), is because these type parameters (TypeVars) are invariant.
If they weren't invariant, thenx: int16 = int8(3) would be a valid statement, which we probably don't want. So the invariance means thatx: number[NBitBase] = int8(3) is rejected, butx: number[Any] = int8(3) is accepted.

I usually tend to avoid usingAny, but because those_NBit type-parameters aren't actually annotating anything, theAny won't affect anything "outside" of thenumber types, so it's no problem to use it here.