Abstract
AI techniques were introduced long ago in the field of government and regulation, and are now being deployed at high speed in all branches of law, considerably transforming the tools and logic of administrative and judicial action. This development is in line with Leibniz's 17th-century vision of a mathematized, calculable law, but it also involves techniques, such as profiling, which have their origins in the 19th-century normative use of probabilities. Their large-scale deployment, often premature and without adequate control, challenges the foundations of the rule of law, especially the control of powers, the protection of rights and the motivation of decisions affecting them. Despite the fact that several industrial-scale disasters have already been caused by the errors they have provoked, these innovations are part of a regulatory model that is here to stay, and which requires the establishment of appropriate counter-firewalls and safeguards that are also based on technological innovation.
