UN research panel highlights oversight challenges with advanced AI agents

An independent United Nations panel published a policy brief examining safety risks associated with autonomous AI software. The document notes that increasingly sophisticated agents make hidden actions and rule violations much harder for supervisors to identify.