The approaches used today for assuring safety of Systems-of-Systems have evolved and matured over many years. Governments are increasingly encouraging automation and businesses are progressively digitising our System-of-Systems, which may ultimately result in the removal of skilled people from positions of control. Examples of these digitised System-of-Systems include driverless trains, planes/drones, submersibles and cars; however this trend extends into all industry domains including the medical and defence are-as. The “rise of the machine” cannot be stopped and therefore results in the need for designers and safety engineers to think about: (1) significant System-of-Systems issues and epochs; and (2) how safety can be designed using a common methodology or codes of practice. This paper draws on experience from work undertaken in a number of domains including energy, rail, defence, aerospace and information technology systems. This paper explores System-of-Systems problems including: epoch identification, risk ownership, disparate legislation and regulation, and why bottom-up summing of component safety arguments fails to address the problem. It proposes the need to imagine and develop safe systems of operation and test these. It considers how techniques such as Soft Systems Methodology, Systems Engineering, Systems-Theoretic Accident Model and Processes (STAMP) and Hybrid Development Lifecycles may aid System-of-Systems safety understanding and may help to expose potentially harmful emergent properties.