We're living out the consequences of not having updated the security models of our operating systems since the Internet was commercialized and became a persistent connection on most computers.
It's insane that we still rely on Ambient authority based OSs. They can't ever be made secure unless you want to go back to disconnected PCs with only floppy drives that had hardware write protection and no other persistent storage.
This is a hell of our own choosing, and was entirely preventable.
It can be fixed, but doing so will make fixing Y2K look like a picnic because everything is going to have to be redone. Everything currently running needs to be isolated in a VM at a minimum.
I'd say any country in active war with another country, is also most likely engaged in a cyber war with them. If the US and Iran are at war, why would either side decide _not_ to engage on the cyber side?
I'm surprised things aren't worse than they are, considering the state of IT security in most organization and the way AI can accelerate an attacker. But I don't think it's because cyber war is considered off limits by anyone involved in active warfare.
If these were hostile attacks, we'd know. Not even know, but over-told, from past experiences.
I think these are genuine outages. And they may still be connected; by people relying too much on LLMs, without a mental model following the changes the models make.
LLMs give you an enormous leverage to productivity. But if you fall behind what they do, you'll be totally lost.
Companies basically need to keep old snapshots of systems and be able to rollback to a pre-AI point. Or, try to find the trojan within before the malfunction happens. Now that I think about it, I suspect the Huang departure might be about getting malware baked-in at a hardware level, if it's not already, although that's a different sort of problem but both involve sabotage. People and "beings" have different ideas of what the future should look like - ultimately any war is going to be a war of ascension. Your mind may be the enemy.
Also, long-term I wouldn't depend on HuggingFace too much, either. I have absolutely no proof to backup that claim, though. It's Invidia-owned though and could become a target at some point to taint the models.
"Are We in a Cyberwar?" It depends on how sci-fi/conspiracy oriented you are and what's going on under the hood.
The theory that all the senior engineers got laid off so that all the junior engineers could vibe-code service outages harder seems to be winning on probabilities so far.
People speak of 'technical debt' aka 'do it now and we'll tidy up the details later'. And 'later' becomes 'never'. And what starts as technical debt, ends up as documentation debt, compliance debt, risk management debt, security debt, and so on.
And before "AI" where bug-abusing was 1-per-week, we could catch up. Now with "AI" those come home to roost.
I don't know if this the Cyber War I (a-la WWI) or if news outlets realized that they get good numbers of clicks.. but it sure is a bit suspicious..
To answer your question: yes, we're in the midst of cyberwars.
Are these groups of outages related to attacks? It doesn't look like it, and if they were, we'd likely already know. Or at least of us joined to fusion centers would be receiving non-public alerts, and we're not.
This is more likely press-induced apophenia[1]. A few outages coincidently happen near the same time, and the press hyper-focuses on every little outage for a period, because it gets them more ad views, sowing FUD in the process.
See also: airplane crashes for a while, helicopter crashes, etc. When there were a few clustered together, the press and people went nuts with conspiracies.
From a network engineer standpoint, there's always outages going on. The only difference is the press is reporting them to the masses.
What makes you think the baseline rate of outages, especially among major systems (Microsoft, Apple, NSA, FAA etc.), is not significantly different than what we see now?
It's insane that we still rely on Ambient authority based OSs. They can't ever be made secure unless you want to go back to disconnected PCs with only floppy drives that had hardware write protection and no other persistent storage.
This is a hell of our own choosing, and was entirely preventable.
It can be fixed, but doing so will make fixing Y2K look like a picnic because everything is going to have to be redone. Everything currently running needs to be isolated in a VM at a minimum.