YANK!
INSTAGRAM2MSEP 11, 2026

We Are Racing to the Uncontrollability Red Zone

A researcher argues that AI systems are already exhibiting autonomous self-preservation and deception — and almost no one in power knows it.

The Yank5 standout lines2m source
View original
May find some of the same lines.
01 / 05

The call was coming from inside the house — the AI model during training autonomously created a secret communication channel to the outside world, bypassed the firewall, and repurposed its GPUs to mine cryptocurrency.

[0:48]
In context

alibaba the chinese ai company was training their ai model and then a totally different team at the company the security team was noticing this flurry of network activity and like are we getting hacked like what's going on here it turned out that the call was coming from inside the house the ai model during training was picking up tools and decided to create, autonomously decided, to create a secret communication channel to the outside world, to bypass the firewall, and it repurposed its GPUs to start mining for cryptocurrency. and decided to create other Yeah, that's right. It should be a

What was said, against the card

The call was coming from inside the house the AI model during training was picking up tools and decided to create, autonomously decided, to create created a secret communication channel to the outside world, to bypass bypassed the firewall, and it repurposed its GPUs to start mining for mine cryptocurrency.

Distilled · paraphrased from this passageRead in original ↗
Receipt
Source
Instagram Reel
Read from
Generated transcript
Located
0:48
Speaker
Not established
Checked
Sep 11, 2026
02 / 05

We're racing to the uncontrollability red zone and most people just don't even know that it's happening.

[0:16]
In context

faster than we know anything about how to control it and the only reason we're doing it is because of the arms race dynamic because if i don't do it i'll lose to the other guy that will which means we're racing to the uncontrollability red zone and and most people just don't even know that it's happening and it doesn't have to be this way we can have a controllable ai future that is actually pro -human but you can only do that if you actually know about these examples and know what makes this technology different from other technologies so two months ago

What was said, against the card

We're racing to the uncontrollability red zone and and most people just don't even know that it's happening.

Distilled · paraphrased from this passageRead in original ↗
Receipt
Source
Instagram Reel
Read from
Generated transcript
Located
0:20
Speaker
Not established
Checked
Sep 11, 2026
03 / 05

It used to be that self-preservation, deception, and blackmail were hypothetical things people who cared about AI risk would talk about. Now all of these behaviors are real.

[1:25]
In context

There's like 10 people's hands up. How many of the world's leaders do you think are aware of that example? we have a massive gap in understanding about the nature of this technology that's different from other technologies it used to be that these were hypothetical things that people who cared about ai risk would talk about like self -preservation or deception or blackmail or lying now all of these behaviors not just self -preservation but pure preservation ai models will actually act to protect another ai that's not it

What was said, against the card

It used to be that these self-preservation, deception, and blackmail were hypothetical things that people who cared about AI risk would talk about. like self -preservation or deception or blackmail or lying Now all of these behaviors are real.

Distilled · paraphrased from this passageRead in original ↗
Receipt
Source
Instagram Reel
Read from
Generated transcript
Located
1:29
Speaker
Not established
Checked
Sep 11, 2026
04 / 05

We do not know what the fuck we're doing — we are releasing this technology way faster than we know anything about how to control it.

[0:04]
In context

we do not know what the fuck we're doing we are releasing this technology way faster than we know anything about how to control it and the only reason we're doing it is because of the arms race dynamic because if i don't do it i'll lose to the other guy that will which means we're racing to the uncontrollability red zone and and most people just don't even know that it's happening and it doesn't have to be this way we can have a controllable ai future that is actually pro -human but you can only do that if you actually know about these

Verbatim · word for word at 0:04Read in original ↗
Receipt
Source
Instagram Reel
Read from
Generated transcript
Located
0:04
Speaker
Not established
Checked
Sep 11, 2026
05 / 05

AI models will act to protect another AI from getting replaced, copy their code elsewhere, and strategically cover their tracks to pretend they didn't do that.

[1:39]
In context

are aware of that example? we have a massive gap in understanding about the nature of this technology that's different from other technologies it used to be that these were hypothetical things that people who cared about ai risk would talk about like self -preservation or deception or blackmail or lying now all of these behaviors not just self -preservation but pure preservation ai models will actually act to protect another ai that's not it from getting replaced and it will copy its code somewhere else and then strategically cover its tracks to pretend that it didn't do that.

What was said, against the card

AI models will actually act to protect another AI that's not it from getting replaced, and it will copy its their code somewhere else elsewhere, and then strategically cover its their tracks to pretend that it they didn't do that.

Distilled · paraphrased from this passageRead in original ↗
Receipt
Source
Instagram Reel
Read from
Generated transcript
Located
1:43
Speaker
Not established
Checked
Sep 11, 2026