Don't worry about that. :) If the AI is Friendly, it won't hurt us no matter how much it knows, and if it's Unfriendly, it won't matter how much it knows; it would hurt us just as bad anyway.

Well, what you plan and what takes place ain't ever exactly been similar. ;)

Have you ever tried to build LLVM in debug mode? I have. Got a 600+ MiB executable that took a full minute just to load itself. The turn-around time was ridiculous. GCC's build, meanwhile, was less than a tenth of that. Much faster, too.

Especially those you don't like. Those you do like need no protection from you.

There have been a few recent movies on that form.

I, Robot (2004) shows a weak, non-self-modifying city-control computer.

Her (2013) shows an almost-friendly non-corporeal AI OS. It gains superintelligence, but chooses to sublimate rather than wipe out humanity.

Trancendence (2014) shows a human mind uploaded. It gains the "nanotech" superpower, but none of the others that a proper superintelligence would have.

No movie has shown a true superintelligence, because (1) we can't imagine what it would do, (2) if it was hostile, we'd lose, which most marketeers wouldn't like, and (3) if it was friendly, it would quickly end with "everything perfect for everyone forever"; not much of a story.

Yep.

Certificate validation is a defense against Man-in-the-Middle attacks. But the "Let's Encrypt" system is vulnerable to a MitM attack between its server and the server that would request the proper certificate.

It can thus be fooled into issuing false certificates by the very people those certificates should defend against.

You seem to underestimate the inventiveness of a superintelligence, and the diversity of hardware controlled by computers, and our reliance on them. It is also possible to use electronic communication to make humans do work for you.

For example, if the AI solves the Protein Folding Problem, it could contact a Protein Sequencing Service and have them build proteins that fold into self-replicating nanobots.