Ex-intel official discusses the risks and opportunities presented by revolutions in AI

"You can defend well hundreds of times, but they only have to get through once," the former official noted. "It is always harder to protect than it is to attack."