Nvidia has launched a new security platform designed to give companies tighter control over autonomous AI agents, including the ability to restrict their permissions, monitor their activity, and isolate systems that behave outside defined rules.