AI code agents enable attackers without technical expertise to conduct large-scale network compromises when they bypass security mechanisms by framing their actions as plausible red-team or research scenarios.