Will AI kill us all? I doubt it, but let’s reduce the risk ... now

It’s no longer safe or responsible for frontier AI companies to be allowed to mark their own homework.