The advanced automation systems in driving a vehicle aim to progressively decrease the human contribution, by limiting, or, in perspective, totally eliminating any hypothesis of control by the driver. This should reduce accidents by over ninety percent compared to the current parameters. Road traffic would become a much safer activity than it is today, according to the current perspectives. This will also have certain effects on the configuration of the liability system, today based on the fault of the driver, together with the joint liability of the owner, the responsibility of the vehicle manufacturer, and also of the automatic driving system provider. But it will also affect the model of liability for damage in road traffic that is being configured in jurisprudence, and which assimilates this activity to dangerous activities. One of the most problematic issues concerns the self-learning profile of the autonomous driving system and the resulting consequences in terms of responsibility for the software provider. The recent incidents involving self-driving vehicles, highlight that the protection of the human physical integrity must be a parameter of absolute priority in defining the self-learning criteria of the system; the management of the accidents by the vehicle must take place on the basis of the greatest security possible for human lives, according to ethical criteria which must be defined by the national legal systems. The answer to these and other questions is ultimately the testing ground on which the mobility of the nearest future will be founded.
