The UK Information Commissioner’s Office (ICO) is two years into its privacy sandbox project while Norway’s data regulator Datatilsynet has just started an Artificial Intelligence (AI) sandbox. Do such safe spaces, where developers and regulators cooperate during the development of products, benefit both sides?(1) And what is the industry perspective on their potential place in the AI regulatory landscape? These questions, and the wider issue of how AI rules and other public policy is developed in the first place, were discussed in an online session at the PL&B 34th Annual International ­Conference on 7 July.