Time to start taking AI risks seriously

Governments promise a light-touch regulatory agenda amid growing threats