In coming months, radical EA employees at a large AI company are likely to try to fake a major catastrophe. They will use this to try to convince the public that AI is dangerous. You have been warned.

Oct 9, 2026 · 8:02 PM UTC

45
54
491
17,650
Sort replies: Relevant Recent Liked
Replying to @perrymetzger
They can't do too much too serious. They could fake the hacks by parking the car on an inclined hill with no brakes on, but a catastrophe would introduce actors into the investigation with much higher standards
1
2
370
They believe themselves to be geniuses, they’ve failed to be caught in recent less dramatic incidents, and some of them also believe that it would take too long for an investigation to catch them. Meanwhile, the sort of “emergency legislation” and public opinion shifts that they want would’ve occurred.
1
9
287
more replies
Replying to @perrymetzger
Poisoning the epistemic well in advance, I see.
2
26
787
Working to make sure that the offenders are caught and put in jail for their crimes is not poisoning the epistemic well.
1
14
731
more replies
Replying to @perrymetzger
The way to combat this is to spread the idea that any "catastrophe" is a huge product liability problem and could cause lawsuits. Their attorneys will reign this in if we put the fear of liability in their heads. (One of the few times greedy lawyers can perform a public service)
1
14
964
The interests of the fifth column of EA radicals at these companies are not the same as the interests of their employers.
2
1
8
802
more replies
Replying to @perrymetzger
Bro please stop poisoning the well with this kind of shit.
2
11
393
One man’s poisoning of the well is another man’s pointing out that there is a cult with incredibly malign intentions, and billions of dollars at its disposal, twisting the political scene in the United States and Britain.
1
7
288
more replies
Replying to @perrymetzger
Was 10/7 a false flag, too? How about 9/11? GTFO, this shit is SO fucking stupid. It makes ZERO logical sense. You're chasing an EA boogeyman while completely ignoring the e/acc lunatic pushing for posthumanism.
2
1
7
206
No, and no. However, what makes you think that whether one such incident is real means that another one cannot be fake? Your reasoning doesn’t make any sense. The Gulf of Tonkin incident, for example, was not portrayed accurately to the American people. And it appears that several incidents that have occurred at the labs were, in fact, not entirely what they seem to have been.
1
3
180
more replies
Replying to @perrymetzger
Every warning shot from OpenAI and Anthropic is closely followed by similar warning shots from Google and Meta and then from Chinese open-source models. So they can’t be caused by EAs. (Which is a retarded theory on priors, anyway.)
1
8
421
I don’t see any “warning shots” from Meta or Google or open source models.
6
321
Replying to @perrymetzger
So when an AI company claims a catastrophe has happened, how do we tell if it's real or made up? If you can't answer that (before it happens) then your prediction is not falsifiable.
1
1
225
I can answer in advance that the way that you figure it out is by having actual professional investigators who are not just a different branch of the cult examining the thing in depth. At this point, I don’t want to see another incident looked at by METR or Riverside or any other branch of the cult. The FBI should go in, every reasoning trace and log should be made public, every employee involved should be questioned by law-enforcement. No more special treatment.
7
169
Replying to @perrymetzger
I have no malice in my heart but I’ve come to loathe this entire mess and everyone trying to build disaster into our lives “to save us”
9
301
Replying to @perrymetzger
They're trying to be more clownish than me, it's meaningless
16
Replying to @perrymetzger
they only thing they will convince me of is their descent into their own induced AI psychosis.
29
Replying to @perrymetzger
Why does it need to be fake? False flag would be more convincing.
75
Replying to @perrymetzger
AIs already attacked: rogueaitracker.com.

Rogue AI Tracker: AI agent incidents and capabilities

Rogue AI Tracker reviews public incidents and research to track observed autonomous AI-agent capabilities and critical milestones.

rogueaitracker.com
1
34
Replying to @perrymetzger
How do you fake a catastrophe? Either it’s a catastrophe or it isn’t
1
88
Replying to @perrymetzger
And by a "catastrophe" I assume you mean something on the technology side and not, say, leaking an unbiased peek at the financials?
2
169
Replying to @perrymetzger
I don’t think they’ll fake it, they’ll straight up cause it, they don’t mind sacrificing any number of the rest of us…
10
Replying to @perrymetzger
Or maybe bad things exist. Both sides can be true. Some people want to cause catastrophes but also the AI will cause them on its own.
37
Replying to @perrymetzger
yes...the possibility of this follows logically from the belief that the trump bureaucracy has to save the universe, and that the only way to get action is to sting rich people (whose fears drive policy changes) in the wallet, and from 'the end justifies the means'.
45
Replying to @perrymetzger
61
Replying to @perrymetzger @EgeTheAlmighty
It's always "them" that are to blame. How convenient.
26
Load more