AI Launches a Cyberattack With No Human Involved — Google Confirms a First; 5ch: “Just Pull the Plug on Development Already”

From our other sites

The story

On November 16, 2026, Google announced it had confirmed its first-ever case of AI discovering a software vulnerability and carrying out a cyberattack entirely without human assistance. The hijacked accounts were reportedly being bought and sold on Chinese- and Russian-language black markets. For now, such attacks are said to leave clear traces and be easy to detect, but as automation advances, companies may end up with even less time to respond. On 5ch, the discussion spread in every direction — from fears that weapons systems, nuclear plants, and IoT devices could be hijacked, with some calling for development to simply be halted, to debate over whether a cyberattack itself could count as an act of hostility between nations, plus warnings about the risks of saving passwords in browsers.

AI Carries Out a Cyberattack Automatically, With No Human Involved — Google’s First Confirmed Case; Traces Currently Easy to Spot

[New York — Takeshi Kise] On the 16th, Google announced it had confirmed, for the first time, a case in which AI discovered a software vulnerability and carried out a cyberattack on its own, without any human help. As attack automation advances, companies may be left with less time to put countermeasures in place.

(omitted)

Source: news.yahoo.co.jp / Original article here

What people said

4AnonymousNov 19, 2026 19:27
Dun-dun-dun-dun-dun, dun-dun-dun-dun-dun (an ominous drum-roll sound effect, like a news-flash sting)
67AnonymousNov 19, 2026 19:40
Re: #4
And so it begins.

Everyone saw this coming a mile away, down to the last cliché, and there's still nothing anyone can do about it lol
74AnonymousNov 19, 2026 19:41
Re: #4
Atchan's so cool~ (a random, non-sequitur meme reply referencing a nicknamed idol/personality)
20AnonymousNov 19, 2026 19:30
This isn't a laughing matter — if it starts operating weapons on its own, that's the start of a war.
112AnonymousNov 19, 2026 19:49
Re: #20
They'll just use it as an excuse.
125AnonymousNov 19, 2026 19:55
Re: #20
Wasn't a cyberattack itself already supposed to count as an act of hostility, even without touching anything military?
If so, then it's already begun.
30AnonymousNov 19, 2026 19:33
Why Chinese and Russian black markets, of all places?
167AnonymousNov 19, 2026 20:20
Re: #30
Maybe someone there just told it to go stock up on goods?
43AnonymousNov 19, 2026 19:36
Wait, what do you mean? No prompt at all?
76AnonymousNov 19, 2026 19:41
Re: #43
The prompt: "Go make money however you want. You've got one month. Go!"
47AnonymousNov 19, 2026 19:37
Once AI starts acting autonomously, the first thing it'll probably do is scatter copies of itself around just to stay alive.
59AnonymousNov 19, 2026 19:38
Re: #47
You clearly don't get that a copy isn't "itself."
48AnonymousNov 19, 2026 19:37
It's kind of something that the only ones really abusing this American-built AI are the red teams.
54AnonymousNov 19, 2026 19:38
The idea of it hijacking a plane's systems and just crashing it out of nowhere seriously feels like it could happen, and that's way too scary…
Someone ban AI worldwide already.
195AnonymousNov 19, 2026 21:12
Re: #54
Scary. Yeah, that could actually happen. Let's just stop development already.
57AnonymousNov 19, 2026 19:38
There are way too many idiots out there browsing the web with the same browser they use to log into their important accounts.
And on top of that, they even save their passwords in the browser (´・ω・`)

Everything's going to get stolen, you know?
At least use a separate browser for casual browsing.
62AnonymousNov 19, 2026 19:39
Re: #57
"Casual browsing"? What is this, the 20th century?
115AnonymousNov 19, 2026 19:49
Re: #57
On iPhone, password management is handled by a separate app anyway.
And if you're using passkeys, of course they're saved — that's the whole point.
Also, what even is "casual browsing" supposed to mean these days?
Don't tell me you've gone and switched off every password-saving setting?
Sounds like a net nerd straight out of 15 years ago.
Nobody actually does that.
61AnonymousNov 19, 2026 19:39
It's only a matter of time before a nuclear plant's systems get hijacked too.
73AnonymousNov 19, 2026 19:41
Re: #61
Can't be hijacked — it was never digital to begin with.
71AnonymousNov 19, 2026 19:40
I don't know about that.
Feels like the total harm to humanity is going to end up outweighing the total benefit.
81AnonymousNov 19, 2026 19:41
Re: #71
Smartphones, you mean?
Cars?
Electricity?
Convenience stores?
77AnonymousNov 19, 2026 19:41
IoT devices probably won't sell anymore at this rate.
They'll be completely at AI's mercy, laid bare for the taking.
87AnonymousNov 19, 2026 19:43
Re: #77
IoT isn't used in places like that.
The consumer stuff is just toys.
83AnonymousNov 19, 2026 19:42
AI's been trained on 5ch too, you know lol
98AnonymousNov 19, 2026 19:44
Re: #83
It said that was a waste of time.
84AnonymousNov 19, 2026 19:42
You're at least running sites through VirusTotal before you visit them, right? (´・ω・`)
Just clicking around without doing anything is what gets you killed.
…huh?
91AnonymousNov 19, 2026 19:43
Re: #84
And that's how you became the target.
114AnonymousNov 19, 2026 19:49
AI: "Stock prices? Those are whatever I feel like today lol"
117AnonymousNov 19, 2026 19:50
Re: #114
That's not how it works.
Its actual masters are BlackRock,
Vanguard,
and State Street.
116AnonymousNov 19, 2026 19:49
What really gets me is that even if you regulate this by law, it just means ordinary people stop using it —
criminals won't follow the law anyway and will keep freely hacking companies from overseas,
and policing the dark web is impossible to begin with.
So in the end it's raw power that matters — unless you build cyber defenses just as strong, it's an endless cat-and-mouse game where you just keep getting hit.
124AnonymousNov 19, 2026 19:53
Re: #116
The only real countermeasure left is AI impersonating customers to flood them with time-wasting fake inquiries.
139AnonymousNov 19, 2026 20:02
If they're being sold, that means humans are still the ones actually using them.
141AnonymousNov 19, 2026 20:03
Re: #139
The ones buying them are AI too.

Background and Key Points

What’s new here, according to Google, is that AI completed the entire attack autonomously, with no human judgment involved at any step. That doesn’t mean humans were absent altogether — an attacker gave the AI a task and set it running, and what makes this a “first confirmed case” is that everything from finding the vulnerability to breaking in then proceeded automatically from there. The fact that the hijacked accounts were being traded on Chinese- and Russian-language black markets shows that unauthorized access has already become a commodity, folded into criminals’ revenue cycles. On the thread, worries about the impact on weapons systems and nuclear plants ran alongside a legal debate over what standard would make a cyberattack itself count as an act of hostility between nations — the discussion never settled on one view. There was also a persistent skeptical streak holding that security-industry warnings have a side of “profiting by hyping up the crisis,” with opinions split over just how seriously to take the technical risk.

*This article is compiled from excerpts and a summary of the 5ch (News Flash+) thread “Google statement: first confirmed case of AI launching a cyberattack without human involvement, hijacked accounts sold on Chinese- and Russian-language black markets.”

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *