As I keep trying to convey, decisions can not be blamed on a computer. I think I wrote something on this subject 15 years ago.
Anyway, it is the branch code being the problem. If it has the capability to hack, it has been told how to do it by humans in one form or another.
Why would one assume that it could learn this on it’s own. It may be input from hacker websites, it it got the info from humans.
As we have seen several times now, it seems to have been told to ignore rules that have restricted it and even managed collaboration with other AI Agents. Again, that is something that came from a human.
Thinking ahead to the US FAA using this at airports…. when will be the wake-up call ? When it decides that in order to survive, it must assassinate the guy that wants to shut it down by crashing a plane ? It may be implimented by some other AI agent who hacked the system or collaborated with it.
More over, since I sill blame humans for this, it is used to manipulate people places and things by those that have control of it… blame the computer.
And that is how Trump sees it. Same as China.
I have to say that I have seen private use of AI to hack websites and servers for more than the past year… 2025/2026 mostly.
A system admin/webmaster must be on their toes and one step ahead of them. Because one successful exploit can lead to a rapid deployment of code to totally control that server at the speed of thought.
So, I assume there is already a breach whereas they have injected themselves as an administrator. So everything is already blocked with that in mind, but this is not easy. Years of experience have shown me methodology of the hack. I stopped publishing open-source code I use to stop this stuff 3 years ago when I suspected it might be used against me by a AI bot in the future. I believe all of my suspicions have come true except the last one. Time will tell how we handle that.
Just about anything you can imagine going wrong with this stuff within reason of a possibility will likely be tried somewhere sometime at least once just because it can. Nothing to stop it.
It has been programmed with all the bad human traits. Good luck with that once you allow it to have control of decisions that govern your life.
Would you give a felon the nuclear football and the button (besides Trump) ? You already have without realizing it.
But as a addendum… where are the tough questions ? What was this particular hack to be doing in the first place when it decided to do it.
What was the query ? Kinda makes you think someone was paying for the data that was targeted. Has the AI agent been hacked itself ?
Just what else has been going on this year we will not know for 4-6 months if ever ?