Can We Regulate AI to Prevent Harm?

This talk examines why regulating AI currently appears unfeasible at both the domestic level and—given the current lack of consensus—at the international level. After further exploring the limitations of (tort) liability and other private law mechanisms for accountability, it looks at the (again limited) extent to which harmful uses could be excluded through technology – i.e. built into AI applications and AI services. In light of the limits of these different means to prevent or mitigate harm, the talk appeals to AI developers and AI users to take their responsibility for the tech they develop, deploy and apply seriously.