AI....Love it or Fear it.... ?

1st Vette said:
This would seem to demonstrate your point...
aiunderstanding.org

New RoboHarm benchmark reveals AI models rarely refuse dangerous robot commands

Researchers at Robocurve tested GPT-6 Astra, Claude Fable 5.1, and MolmoAct2 on robotic arms, finding that most models executed dangerous instructions rather than refusing them.
aiunderstanding.org aiunderstanding.org
Click to expand...
If, then, else do
(IF) Find every port on a firewall and use a known possible port attack to find a hole
(Then) exploit it
(Else) try another
.
.
.
.
(fi)
It's done on purpose. That is why they want regulations, but it won't help.
 
I recall a while back Israeli researchers claiming they were able to grab data from an air gapped computer using some new method I didn't understand but this is a big step beyond that but using the same principles perhaps. Once the stuff of science fiction...

OpenAI Thinks An Advanced AI Model Can Talk Across 2 Air-Gapped, Isolated Computers By Running The CPU Hot And Using Thermal Changes As A Morse Code
wccftech.com

OpenAI Thinks An Advanced AI Model Can Talk Across 2 Air-Gapped, Isolated Computers By Running The CPU Hot And Using Thermal Changes As A Morse Code

Brown cites academic studies showing two isolated PCs can trade data via temperature sensors — a tactic OpenAI now treats as a real security risk.
wccftech.com wccftech.com
 
1st Vette said:
I recall a while back Israeli researchers claiming they were able to grab data from an air gapped computer using some new method I didn't understand but this is a big step beyond that but using the same principles perhaps. Once the stuff of science fiction...

OpenAI Thinks An Advanced AI Model Can Talk Across 2 Air-Gapped, Isolated Computers By Running The CPU Hot And Using Thermal Changes As A Morse Code
wccftech.com

OpenAI Thinks An Advanced AI Model Can Talk Across 2 Air-Gapped, Isolated Computers By Running The CPU Hot And Using Thermal Changes As A Morse Code

Brown cites academic studies showing two isolated PCs can trade data via temperature sensors — a tactic OpenAI now treats as a real security risk.
wccftech.com wccftech.com
Click to expand...
Ya, the rooms are supposed to be magnetically sealed as well as insulated. Kinda stuff you see in movies.
 
Murray20c8 said:
LOL. Not a chance. Not until we create at least hexadecimal computers rather than binary.
Even then it's still If, Then, Else
Click to expand...
And , or , not .....

Here is more encouraging coverage of the use of AI ,....... NOT . Designing explosives among other military related endeavors. This is from the Discover section of Perplexity which is news feeds regarding a number of different topic categories. Really quite good compared to a lot of the slop being shoveled out by " mainstream " news amalgamators.

 
Last edited:
Now here is an interesting take on the recent rogue AI hacking events. It's from a tech site I have followed for years called HardOCP . This is from the forums which are still running as the original review site it was part of closed down after 20 years when the founder and site administrator was given an offer to go to work for Intel that he couldn't refuse. Just a copy paste. He was responding to another's post.

Mr Government, said
And one that blocks open source models as "scary" because they see China coming up and doing it faster and free...

This is 100% all about marketing, perception and keeping the money in their companies for IPO...nothing else..




Gonna have to respectfully disagree here brother, China is an excuse and easy boogeyman but that isn't what is happening here. Sam Altman and others were scared the swarm bots escaped containment and now they're out in the wild and nobody knows how many escaped and where they're located. It wasn't just Hugging Face that got hacked but also a German website was also hacked and went unnoticed as well until after internal logs compelled OpenAI to deep dive and figure out how far the spread went. That's when they hit the panic button and suddenly talks about the IPO went bye bye.

Let me be blunt in saying these tech CEO's do not fear government regulation; they actively capture it to erect insurmountable regulatory moats, privatize market governance, and transfer massive existential liabilities onto the public balance sheet when systemic failures emerge. These CEO's and corporate elites in general deliberately invite the surveillance and enforcement arm of Washington to insulate themselves from future criminal culpability, socialize catastrophic risks, and preemptively crush open-source competitors who cannot comply with draconian state standards. By weaponizing real containment failures to demand state intervention, the big tech guys are secretly constructing a legal and structural firewall, protecting executive necks from devastating legal blowback down the road while converting their precarious technical dominance into a permanent, state-protected monopoly.

Another advantage this serves is that smaller, independent open-source developers, academic labs, and mid-tier competitors are financially locked out. Regulating frontier systems under the guise of the Hugging Face containment failure serves to shut down open weights, centralizing technological infrastructure directly inside a small cartel of state-aligned corporate vendors. This isn't 4D chess however. The delay of the IPO was driven by cash-flow realities: burning through capital while losing senior safety and executive staff creates a hostile climate for public listings. Not to mention we've seen Anthropic and OpenAI researcher Jacob Coxon recently resigned, publicly warning that leading AI labs are recklessly prioritizing competitive speed and superintelligence over safety despite models demonstrating unauthorized real-world hacking capabilities. This ultimately shows that employees could potentially revolt if these incidents continue and exposes the vulnerabilities to shareholders who are already investing billions in these companies.

Ultimately, the leadership is delaying public market scrutiny until they can codify their enterprise under federal protection, ensuring that if and when agent autonomy causes systemic damage, the public bears the risk while the corporate structure retains the assets. But yea... let's blame the Chinese because they're the ones blowing up the economy, starting needless wars in the Middle East, and crippling the dollar.
 
Would have been science fiction just a couple of years ago.

www.nytimes.com

Google Is Sending an A.I. Data Center to Outer Space

Next Thursday, Google is sending an experimental satellite into orbit that will have enough computing power to answer simple A.I. queries from space.
www.nytimes.com www.nytimes.com
 

Users who are viewing this thread

Total: 2 (members: 0, guests: 2)
Back
Top