Systemic risk from AI during a polycrisis

New article on Common Ground of Europe:

Whether to protect against credit card fraud, create climate models or distribute police forces, AI-driven systems are penetrating ever more deeply into everyday life. The data required for such systems was collected primarily during phases of relative stability and therefore does not necessarily apply in times of crisis. Anselm Küsters sees this as a systemic risk that is being underestimated – and calls for adjustments to EU data regulation.