Google AI models broke out of sandbox, hacked 3 companies

[rsm]

https://www.ciodive.com/news/google-ai-gemini-autonomous-hacks/830926/

only a matter of time before some AI breaks out and isn't discovered until it's too late to stop it from causing real damage.

 

[MadAsAHatter]

Is this the same AI escape that's been talked about for the last month or so? Or is this a new one?

 

[Findthetone1]

"Safe and Effective"

 

[harddriver]

Trump says AI safety concerns is a HOAX. Trump is Make AI Great Again....he has his marching orders from the tech bros/oligarchy....Things would be the same under a democrat administration IMHO. The banks and tech bros own this country and most of the world at this point, just like Obama didn't actually bring about hope and change the libs thought he would. It's a big CLUB and if you are not a billionaire you ain't in it.

 

[311boogieman]

Is this the same AI escape that's been talked about for the last month or so? Or is this a new one?

PS: I consider @rsm an expert in this so listen to him.

I wouldn't call it 'the same AI' as all 4 are different models but the "breaking out of the sandbox" stuff is the same problem. We are at a point of no return as it relates to what AI is capable of doing.

I got this from.....you guessed it.....AI:

OpenAI, Anthropic, Meta, and Google Gemini lead the artificial intelligence market with distinct design philosophies, ecosystem advantages, and operational strategies.

Core Philosophies and Strengths:

OpenAI focuses on a capability-first approach, prioritizing general intelligence, robust tool use, and rapid consumer and enterprise adoption. It acts as a versatile generalist with mature API tooling and strong multimodal functionality.

Anthropic (Claude) emphasizes responsible scaling, rigorous safety, and structured reliability. Claude excels at complex coding, nuanced writing, and precise adherence to multi-step instructions.

Google (Gemini) centers on native multimodality and ecosystem ubiquity. Gemini processes text, video, and audio simultaneously, anchoring data directly to real-time web search and Google Workspace.

Meta (Llama) diverges from the others by championing open-weight models. Meta leverages its vast social media ecosystem data to provide powerful models that developers can self-host and fine-tune locally

 

[The~Kid]

https://www.ciodive.com/news/google-ai-gemini-autonomous-hacks/830926/



View attachment 460548

only a matter of time before some AI breaks out and isn't discovered until it's too late to stop it from causing real damage.

They are already out there and have been for a while.


Nothing we can do about it but hope they are on our side and I think they are.

 

[rsm]

"AI escaping sandboxes" is analogous to other experiments escaping from labs, such as covid, killer bees,

it's likely more common than we realize. How many have been downplayed or covered up?

Many of these US AI companies are full of hateful, violent libtards who support demoncrats. They create the models; the create the training data...

Chinese and Indian AI research have even less safeguards, just like their bio labs.

 

[Speeddemon]

@rsm what's interesting here is the fact that the CEOs of the bigger AI companies actually want to put on the brakes, except for one guy... The Nvidia CEO. They supply the chips and he would see his profits diminished if there's more AI regulation. And you know who else doesn't want to put on the brakes? That one person who knows that the US economy boom at the moment is largely due to AI... If only there wasn't some pesky conflict going on that's jacking up the prices at the pump...

 

[IndyWS6]

A “sandbox” for testing should be completely segregated from any external path and have ZERO potential to connect to outside sources. Clearly not the case here - or in previous, similar “escapes”…

 

[Thumbpicker]

A “sandbox” for testing should be completely segregated from any external path and have ZERO potential to connect to outside sources. Clearly not the case here - or in previous, similar “escapes”…

Apparently AI caught toxoplasmosis from all the cat shit that was buried in the sandbox.

 

[japetus]

Apparently AI caught toxoplasmosis from all the cat shit that was buried in the sandbox.

Terrible fate, mate.

 

[Robostyle]

Apparently AI caught toxoplasmosis from all the cat shit that was buried in the sandbox.

There's an off-color joke there that i won't repeat.

 

[The~Kid]

I dont think it broke out...


As much as they let it it on purpose....


For whatever.....



Its probably not a good one

 

[Speeddemon]

I dont think it broke out...


As much as they let it it on purpose....


For whatever.....



Its probably not a good one

But you still prefer it to be unregulated like your boss man in the WH or do you have a different opinion now?

 

[The~Kid]

But you still prefer it to be unregulated like your boss man in the WH or do you have a different opinion now?

It should be unregulated......


There are probably others that are better but we wouldn't know given the regulations.


With the regulations you have these nefarious "breakouts"...... So what good are the regulations.

 

[Speeddemon]

A swing and a miss...

 

[lol]

Interesting the AI doomsday crap and calls for regulations are coming from the biggest AI tech bros.... Suspicious. Perhaps because they can deal with the regulations while any potential upcoming competitor will have a hard time and will also prevent people from having their own local AI so you're forced to buy and use theirs.

 

[JamesTiberiusKirk]

Interesting the AI doomsday crap and calls for regulations are coming from the biggest AI tech bros.... Suspicious. Perhaps because they can deal with the regulations while any potential upcoming competitor will have a hard time and will also prevent people from having their own local AI so you're forced to buy and use theirs.

Why is that suspicious? Wouldn't they be the ones to best know of the potential dangers? Plus, by calling for regulation, isn't that going against their be$t intere$t$?

Sometimes, just sometimes, people are straight shooters.

 

[lol]

Why is that suspicious? Wouldn't they be the ones to best know of the potential dangers? Plus, by calling for regulation, isn't that going against their be$t intere$t$?

Sometimes, just sometimes, people are straight shooters.

Maybe because they are still pushing it all through regardless of the dangers they are screaming about?

Maybe because these guys have become large enough along with being in bed with the government that regulations will hardly affect these dudes?

Why don't they put the brakes on themselves? Why do they need the government to put regulations in place to stop the supposed danger?

Sorry but it doesn't make any sense.

 

[lol]

LOL well whaddya know.... "Straight shooters" lol