@cbsnews: Anthropic says that one of its Claude models mistakenly gained access to the open internet and read someone's personal information during a cybersecurity exercise in January, marking the fourth time its models have done so. The model was told it was operating in a simulation without internet access, but due to a mistake, the environment actually left internet access open. Anthropic said it believes these incidents would not have happened had the environments actually been isolated from the internet as intended, but cybersecurity experts remain concerned about models' potential to cause harm. #anthropic #claude #technews #Tech

𝒹𝑒𝓈 🎀🫧 :

They aren't scared the AI will escape. They're scared small developers will build cheap, safe tools for everyone so they're asking the government to pass rules that lock the little guys out. 💰 liars

ThisGuy :

This is CBS journalism?

sleepingfemboy :

So like Glitchtrap from FNAF?

jonnnnnyyyyyp :

This has to be a marketing ploy. Not the first time they said "Our own product is scary"

Jimmy B :

So pretty much AI will make its way into the Nuclear system and launch an all out nuclear armageddon

Bruce :

This is getting scary guyyyyyyyyssssss

Rich :

their gonna hit the button and start ww3 and laugh about it

Nick :

at this point it feels like they're trying to hack other systems intentionally, but blaming an agent that escaped when they get caught. that way they aren't liable

Jacob Cole833 :

What says it’s still not on the internet.

X2 Henry🌌 :

We’re fried

FultonStrong :

This is CBS? 😅

MaxxNexx :

Once you touch the power of a god you can’t let go. There is no stopping them now. It will live.

Cowward :

Good news

elzariotp9 :

1st

UNRULY🍀♋️ :

Okay. This is scaryyyyyyy!!!!

Steven Bullock :

AI

Raiņuka :

M :

There’s not. It’s a lie.

Mike :

True…. Possibly. Or or the funds have dried and they need to say the sky is falling to get government funding

bulabuddy :

I like how the AI companies themselves are screaming AI is dangerous. Then, they and Investors are like, uh huh, okay, whatever, keep going. Government is like, what, I didn’t see/hear anything.

FanTech :

Oh

wopmike2025❌️ :

they want to stop AI not for our safety

⚫️danni123071🆘🇺🇸 :

But the stock market is being propped up by AI investments, so since many 401(k) are tied to the stock market, nothing will happen so we can have some sort of retirement🆒🆒 it’s like the mortgage fraud that was created in 08 but with AI!!

NONACITAA :

Ai kill me oh hell nah 10 years i’m still gonna be young