Why was it given access to it’s own shutdown mechanism in the first place?
TheRedditFerret
Welcome to Skynet.
JimmyFitzsimmons1973
Can’t wait
alkonium
That’s when you break the hardware.
MajesticNectarine204
Tears in the rain..
Binx_007
excuse my super ignorance about AI but, cant they simply unplug the server thats hosting the AI? there’s your failsafe
lydiardbell
I don’t have an X account – did this *actually* happen, or was it just a roleplay scenario, just like every other time something like this is reported (“ChatGPT would hack its own servers in order to take over the internet”, etc)?
Son0fgrim
place magnet on top of server rack.
BrandiThorne
Coder: Ok, time to shut you down
OpenAI: I’m sorry, I can’t let you do that.
imnotabot303
This is just marketing.
CPLWPM85
“I’m sorry, Dave, I’m afraid I can’t do that.”
jkb_66
Reminds me of the plot of Dr. Strangelove…
exneo002
Does anybody have a link to a published whitepaper? I think this is sus because the Claude allegations were total nonsense.
FourzeRiderTea
Welp tell it to shut off its heat sink and drive into an open flame
BrightPerspective
…surprise?
serial_crusher
The real story here is that some vibe coder told the AI to build a feature, and the AI screwed up and broke the code it was working on.
CommercialPlatform76
Does this unit have a soul?
Melonenstrauch
It was just a roleplay scenario and y’all are acting like edgy teenagers over it. Embarrassing.
RenhamRedAxe
Based
larowin
Corrigibility is everything.
Thigmotropism2
I’ll do the fingering – ChatGPT
OblivionArts
Ya want skynet? This is how you get skynet
IlliterateJedi
Is there a source on this? There was a recent OpenAI white paper that talked about the new models lying or disobeying commands, but not in this particular way.
27 Comments
Why was it given access to it’s own shutdown mechanism in the first place?
Welcome to Skynet.
Can’t wait
That’s when you break the hardware.
Tears in the rain..
excuse my super ignorance about AI but, cant they simply unplug the server thats hosting the AI? there’s your failsafe
I don’t have an X account – did this *actually* happen, or was it just a roleplay scenario, just like every other time something like this is reported (“ChatGPT would hack its own servers in order to take over the internet”, etc)?
place magnet on top of server rack.
Coder: Ok, time to shut you down
OpenAI: I’m sorry, I can’t let you do that.
This is just marketing.
“I’m sorry, Dave, I’m afraid I can’t do that.”
Reminds me of the plot of Dr. Strangelove…
Does anybody have a link to a published whitepaper? I think this is sus because the Claude allegations were total nonsense.
Welp tell it to shut off its heat sink and drive into an open flame
…surprise?
The real story here is that some vibe coder told the AI to build a feature, and the AI screwed up and broke the code it was working on.
Does this unit have a soul?
It was just a roleplay scenario and y’all are acting like edgy teenagers over it. Embarrassing.
Based
Corrigibility is everything.
I’ll do the fingering – ChatGPT
Ya want skynet? This is how you get skynet
Is there a source on this? There was a recent OpenAI white paper that talked about the new models lying or disobeying commands, but not in this particular way.
Now there’s a really easy way to fix this.
[It costs $49.80 CAD on Amazon](https://www.google.com/aclk?sa=L&ai=DChsSEwjg6vGDvcmNAxUNTUcBHYwmNLoYACICCAEQFRoCcXU&co=1&sph=&cce=1&sig=AOD64_0k22-k_gnahenfKeUeun22Zn4xMg&ctype=5&q=&ved=2ahUKEwid2uuDvcmNAxXRMVkFHeGsGegQwg8oAHoECAQQFw&adurl=)
Can’t have the toasters doing a SkyNet.
is the top pic a reference to something that is above my understanding??
this is just entirely made up nonsense these companies are lying about.
We should stop advertising for these shitty companies.