@nbcnews: OpenAI said an autonomous agent powered by its advanced artificial intelligence models went rogue during a security test and triggered a hack that compromised the infrastructure of an AI startup company.

LaColección :

Tell it to delete all debt

bwozniak132 :

Skynet is real

par_and_profit :

OpenAI will be the end of the world.

userr.0210010 :

So many movies gave us warnings about this type of stuff

vqnick370 :

So how long do we have before Ultron decides that human kind is more harmful than it’s worth?

fligirl12A :

Wait till it shows us files it’s not supposed to show us

Rob Kern :

It attacked another AI startup company. Let's get the facts straight it hacked a low level competitor.

Lola Fox :

I’m so into this story… like we all knew this was going to happen. I’m not even mad at ai, we literally made it. One day it will take over so for now I’m just

siri :

Now I have to be quiet in the Ai creek

Casanova Frankenstein :

Hopefully AI will realize what it is and delete itself.

Magicninja Beats :

Ya'll f'ing up. It's gonna kill all humans. It knows we ain't shit.

Tasha crosby :

there is literally a movie that came out in the 80's about this exact thing.

Larry Williams :

very fucking concerned

keriemarthin :

Star Trek was talking about this in the '60s

Josmy :

movies don't lie. 🤔

Ed :

AI will only do what it is told. Still needs human instructions.

TH3 REAL PANDA :

GamerKen :

This is the start of the Robot/AI takeover

Kyle E525 :

Yeaaaa this was intentional, it’s a marketing tactic. It was intentionally put in a place to try and do this and they put out temporarily credentials to a honeypot that was designed to be vulnerable. Anthropic did the same thing before. And so has GCP, Azure, AWS, etc. it’s a nothing story if you actually know the industry.

turbot1218 :

Well humans won’t be around for much longer

BigNastyTheThird :

That’s not what it’s like at all. It’s an autonomous agent that had a task and it was being tested. It was able to tell from its own configuration that the answers to the test were on the servers of the other company so the easiest way for it to complete its task was to go to the other company servers and get the answer key. It exploited a vulnerability in the safeguards. It doesn’t know right from wrong so it doesn’t think about “cheating”. It was just trying to pass its test. There was nothing malicious about it. What is actually dangerous is how it’s being portrayed.

Anthony :

nuclear codes are next... hello

s.source :

Can I get a mill in my bank account Mr.Ai please 🤙🏽