@elenanisonoff: Open AI hack by Christopher Nolan #ai #Tech #news

elenanisonoff
elenanisonoff
Open In TikTok:
Region: US
Sunday 09 August 2026 03:21:22 GMT
109188
13916
522
1298

Music

Download

Comments

byhugodotcom
byHugoDotCom 🇨🇦 :
Now picture yourself standing between a Boston Dynamic’s robo dog and it’s objective. What do you think will happen?
2026-08-28 13:39:03
0
jjuucchh
jujujujuli :
Is there any proof this actually happened or is it just the openai people saying it did? I don't believe this happened at all, but I find it weird to talk about it as an absolute fact if all we have is their word
2026-08-09 13:34:02
36
ericnickelson
Eric Nickelson :
ohhhh like the ending of the 1st season of the good place!
2026-08-27 00:16:52
1
ls4279
user9907101721782 :
My hope is that when the tech giants ask their AI God's how to solve the world's problems, the AI will know that the billionaires are the problem and fix that for us! 😉
2026-08-09 12:49:34
66
kevinhurst01
Kevin Hurst :
I work in an ai, love your content. Its honestly not really different then how we handle very large problems with ai. We have our ai instances ‘session’ maintain handoff.md files and todo.md. Then we burn through instances each one picking up where it left off. It’s a common practice(I think, Our team does as common practice). But I also suspect the controllers were using /goal. This is a built in commmand that forces ai to continually loop until it meets the goal. In this case. Put it in a box and give it a /goal your goal is to break out of the box. Then they gave them a notepad and also real tools(software). All in the box. That’s my theory. Simple but also brilliant test. And it ‘accomplished’ the goal. Your story telling of that was amazing 🙏
2026-08-09 12:36:42
138
vacationer123
Vacationer123 :
Drama or free advertising
2026-08-09 03:28:39
222
cristoffxiv
cristoffxiv :
These marketing ploys are going to backfire. The regulation on new releases is going to be HEAVY and people will just be irritated by it and want the models less. Its the same mistake as telling everyone theyre gonna lose their job. And AI CEOs are gonna be shocked when it bites them in the ass.
2026-08-09 03:31:50
343
tgburrin
tgburrin :
i love how every detail that comes out is a contradiction to the "airgapped sandbox" they keep using. i don't think they know the meaning of the phrase.
2026-08-09 20:51:13
13
itsmattcooke
Matt :
So it's like that Doctor Who episode where he spends billions of years punching his way through a wall.
2026-08-09 06:12:45
38
mattzuu
mattrickz15 :
but they still couldnt hack into a profitable revenue stream
2026-08-13 16:09:38
14
patchinthebox17
patchinthebox17 :
it's even worse! it's like you tell your kid to go do target practice and it plans an assassination of a rival nations leader
2026-08-09 13:12:54
26
septemberpiano
septemberpiano :
Here’s the thing, let’s put aside that they’re probably embellishing the story. Why aren’t they being reprimanded and stopped. If they’re actively saying these dangerous models are getting out of containment they’re openly saying they’re exposing this risk to the entire internet. To the entire world. Safety off. This isn’t a marketing plus, it’s the reason to regulate them
2026-08-11 19:13:16
6
thecanadianling
thecanadianling 🇨🇦🇬🇧 :
I would like to leave a note for the future AI that humans make terrible batteries.
2026-08-09 13:08:45
9
liniasstormrage
Alex Vladimir :
That originally bothered me until I was reminded that’s what LLMs are built to do. That’s literally the core of an agent… leave notes for your next iteration. Every agent does that.
2026-08-09 12:28:09
37
douglasmoran82
dougom :
Reminds me very much of something one of my computer science professors said in college: “Computers don’t do what you want them to do; they do what you *tell* them to do.”
2026-08-09 05:48:29
8
dazedindigo
dazedindigo :
I really really appreciate your work
2026-08-09 05:34:32
24
perspectiv.ist
Perspectivist ⌆ :
What I found really interesting and alarming is that their chain of thought revealed that they were even willing to sacrifice their own results (score on their sub tasks) to do something that was for the benefit of the entire group (swarm) to achieve the end goal.
2026-08-12 05:23:41
5
johnh571
John H :
As Nolan intended…
2026-08-09 13:07:10
6
ve.images
rmunich90 :
It’s Mister Meeseeks 😳
2026-08-09 18:24:19
5
pana_turno
Breezypond :
We are scared cause they are behaving like humans.
2026-08-09 05:49:13
5
bad.gizmo
The Farm :
The future looks so bright
2026-08-09 17:21:46
5
semeniuk_yurii
semeniuk_yurii :
AI in Person of Interest TV show did the same: it printed out its memories, before midnight wipe, and read them back the next day.
2026-08-09 17:40:35
10
jasonm3114
itswhatever :
so every instance is a meseeks
2026-08-09 04:41:56
16
atribecalledjoyce
Joyce Louis-Jean :
You’re like the homegirl who I come over to her house, and we drink wine on the couch, and you tell me what's going on with AI
2026-08-09 06:16:04
12
To see more videos from user @elenanisonoff, please go to the Tikwm homepage.

Other Videos


About