Oh really? Will it have the ability to run an entire lab robotically to do that? If not, then it won't be the AI doing anything. It will be the people doing it. Its power to do anything in the physical world only exists to the extent humans are willing to grant it. 

If the US doesn't develop it, you can be assured that China and Russia will. US scientists are likely to develop it more quickly. Assuming it is possible, Chinese and Russian scientists, given enough time and resources will develop it eventually. If it is possible, there is no stopping it from happening. Someone will do it. It is pointless to pretend otherwise. 

I live in the physical world. For a computer program to kill me, it has to have power over the physical world and some physical mechanism to do that. So, anyone claiming that AI is going to destroy humanity needs to explain the physical mechanism by which that will happen. This article like every other one I have seen making that argument fails to do that.