Literal negative exponents like 2 ** -1 correctly use math.pow (float), but variables aren't detected:
n = -1
x = 2 ** n # 0.5 in Python
n := -1
x := math.powi(2, n) // 0 in V
Detecting this would need data-flow analysis to know n is negative. Not sure it's worth the complexity since most real code uses literal exponents, but documenting it here in case someone hits it.