Check Point’s Adam Ely says rogue AI agents are already a problem inside ordinary companies, not only at frontier AI labs.