We present new counterexamples, which provide stronger limitations to sums-differences statements than were previously known. The main idea to improve over the original counterexamples by Ruzsa is to consider non-uniform probability measures.