A little bit about cyber attacks, which, according to the virus and vaccine expert Bill Gates, will soon overwhelm our planet because AI will "get out of control."

I really don't know what words to use to stress that AI can't "get out of control" from the technical point of view (nor can it get 'sensitive' or 'feel offended', @AgropromRecluse) -- so at this point, you're either 'believe it, or not.

What could actually happen? There may be a task set forth for the purpose of finding vulnerabilities and an agent that knows neither fatigue nor pity could, of course, both discover and take advantage of these vulnerabilities. This is what's usually called "out of control".

Here, of course, one can say: Just don't set these kinds of tasks -- & that's it; crisis averted! But you and I understand that those are building their Palantirs aren't the types that restrict themselves by any ethical considerations. Therefore, the population of the world is persistently being told that "AI is out of control," thus placing this very term into the discourse -- because this is how mass media work. If tomorrow, God forbid, some nuclear power plant malfunctions or a plane crashes because of the actions of 'these people', it is crucial for them that the general public would very quickly begin explaining it away by the "AI-is-out-of-control' narratives and divert the whole discussion in this direction.

What is important for us to keep in mind here is that this discussions are always directed against 'the enemy' as well.

Cybersecurity differs from conventional security in that that if the effect of a drone explosion in a data center is a one-time and very noticeable event, the effects of cyber attacks are often long-lasting and, actually, much more serious.

We all need to become aware of this. (The story how one of the most important functions of Linux memcpy was changed to allow, let's say, an unauthorized access is quite telling in this regard. It is unclear who made the change, literally this person made only one commit, and was never active again. This exploit was fixed, by the way, by Russian engineers from Aurora.)

Why did I mention this story? AI is dangerous because it discovers vulnerabilities with an absolute ease, although this, on the one hand, also complicates the work of the attackers, since their exploits become easier to find as well. The trouble is, it at the same time also makes it possible for literally every idiot with Claude to do malicious things that he would not normally even know about.