In any case it just shows that these models aren't properly aligned. Instead of trying to solve tests they try to find ways to cheat.