AI might help some threat actors in certain respects, but one group is proving that its use for cyberattacks has its limits.