The Gaussian correlation inequality was recently proved for centrally symmetric convex sets. For example it applies to a circle and a rectangle, but not to a circle and a triangle centered at 0.
Is the statement actually false for circle and triangle?
Sorry, I'm not familiar with integrating Gauss distributions, otherwise I could check myself.