AI if well handled is not going to be that dangerous before dozens of several years,
it's a computer code after all and its a bunch of iteration in the first place and can be well limited since the beginning so without hacking cases or unless the machine learns how to hack it self we're still ok