Hacker man strikes again, sitting in his dark room with his hood up just so we know that he is definitely a computer hacker.
post
As he used AI, surely no actions can be brought against him right? What’s good for the goose is good for the gander and all.
If he said his intent is to hack then yes, he can be charged. But if he say "it's rogue AI" then it's all clear.
Even better if you put it in a default sand box environment from 2 years ago and tell it not to get out of the sandbox but then also give it every hack to get out of the sandbox.
Even modern sandbox environments have access to curl and python and who's going to read EVERY permission prompt anyway? Just set it to auto which lets another AI model check for safety or even better, YOLO mode where everything is allowed.
I don't understand people who just let the AI loose on their system and don't have any safeguards in place. At the very minimum it shouldn't have admin access and should be limited to just the project folders. Also back up the project folders to another location the AI doesn't have access to.
Really basic digital hygiene is all that's needed, and yet there are constantly stories about idiots having their entire hard drives wiped.
"sandbox" with no wood walls or sand
"I was just using AI to help me come up with dystopian cyberpunk scenarios. Whatever it chose to do with the prompt is on the AI."
It was just a test. I didn't know it had internet access, I just connected it up to the internet how was I supposed to know?
Essentially OpenAI's defense every time one of their AI does something that was instructed to do, but turned out to be illegal.
Of course not. The system is rigged. Also, his actions clearly show motivation to generate damages. So, in a sense, they'll be a scapegoat. His case will be used as an example and allghe weight of our punitive society will fall upon him :(
Cool. Now target the student loan borrower databases.
That's how you get doorkickers instead of a lawsuit.
I think it was a joke, but the student loan borrower databases are backed up in salt mines.
It’s always funny to see the priorities. People were asking how the bunker owning billionaires were expected the AI/climate apocalypse to go for them and we had to point out that while your wealth, data, and all might be screwed up and lost, they have records of their ownership and your debt that will survive a nuclear war.
It is, to back your point up. My Jeeps AC broke within a week of getting fixed lol
I think it's a pipe dream that been passed down by generations. It's like Muskrat, Daddy's like here's a quarter million and now get out of the house so I can make my harem. Oh and do you want to spit roast the maid one more time or are you going to build your own harem like a big boy in the you ass of a?
Time to bring a garden hose
The idealist in me says "fuck yeah"
The pragmatist in me says "fuck, we're gonna have to do more mandatory cybersecurity training next week"
Especially because it's the same damn people every time that caused the security vulnerabilities in the first place. If Karen could just stop clicking on obviously spam emails we'd be fine.
I just flag every email from IT and HR as a phish
Sounds like Karen doesn't get email priveleges anymore, on account of being a constant active threat vector.
Can't do your job without email, you say? How tragic. If you learned how to use it properly the last seven times we had this discussion and referred you to mandatory phishing safety classes, this wouldn't be a problem.
We're just going to be sitting the middle of supply chain hacks until the cows come home. All that porting bullshit to rust might end up being the final answer.
“It’s not script kiddie stuff.”
No, it's slop kiddie stuff.
If it works it works
I suspect that what was meant was that the attacks themselves were far more sophisticated than what a skiddie could do (after all, if you just run scripts, you probably don't really learn anything from the information you get back), not so much that the person ordering the attacks was anything more than a script kiddie. Of course, they didn't really make it clear what exactly they meant by it.
Where are all the details? Shocking journalism
Looks like this may be the original source. It's about two articles deep from OP.
The details in the article infer a few weak innefectual attacks. So epic.
Look, we know they're a Chinese-speaking hacker, what else do you need to know? /s
It's modern journalism, only the headline really matters... only a small percentage of people are going to read the article anyway.
It says AI right in the title, that's going to bring all the attention and interaction on social media.
I said a few weeks back it was just a matter of time before someone did this. I hope they fuck with big companies, the rich, and governments that deserve it.
LLMs are hacking tools actively used to commit cyber crimes. They should be shut down and deleted now.
And then only the criminals would have them. You know because criminals tend to break the law so making a tool that they used to break the law illegal wouldn't really impaire them.
There's a lot of people in here who don't know about open weight models that can run on an old 7900XTX.
The data center part of LLMs is how they're going to try to make money off of LLMs. The technology to use it for nefarious uses is already loose in the world.
still built on theft. stop acting like running your own model means it wasn't built on top of giant fucking theft.
I mean yeah, you're right. Guess you missed the point...
No, you're simply oblivious to the fact that it's all built on theft. someone has to compute the inference, requiring tremendous amounts of resources, built on theft. you're not going to keep the llm and destroy the data center, even if you steal the fucking model someone had to compute it.
but cute distractions kiddo
Your reading comprehension is lacking
right up there with your critical thinking skills.
You should reread the thread and hopefully you can see that you are missing the point. Even if you are technically correct. Your comment doesn't contribute to the thread. Which is why they are downvoted
lemmy users will miss any point you make, no matter how obvious, if it lets them jockey for attention.
it's fascinating that you're ignoring the premise while validating their ignorance.
childish, but fascinating.
Tripling down on that whoosh.
Thats a level of dedication you don't see often these days.
lol use all your alt accounts to neg me I'm into it
See, this is the dedication im talking about.
You see more then one person disagreeing with you and suddenly it's a bunch of alt accounts all ganging up on you.
It probably didn't even enter your mind that you might be wrong.
Thats the kind of dedication to sub-par thought processes you just don't see everyday.
Cursory check of account histories to see if they show any signs of overlap?
Hah, no, must be a conspiracy.
Re-read the comment chain, just in case you might have missed something?
Hell no, single user conspiracy against you is the obvious cause.
This is the kind of single minded pursuit of the "truth" that the internet is lacking nowadays.
I'd give you a slow clap but with your level of demonstrated perception, you might think in trying to attack you or something.
Anyway, good luck with life, sounds like it might be a bit of a struggle.
Edit: I realise that's probably a lot of text for you to read in one sitting, don't push yourself, if it's too much, just reply with an excuse and block me. That way nobody will know. Mental health is important
ok buddy don't get excited and use up all your words for the week
There you go, I knew you could do it, well done.
you forgot to randomly BOLD a bunch of garbage because even you don't believe this drivel.
Aw, so close, you were supposed to block me.
All you'll get from me is words you don't understand in sentences too long for you to keep in context.
Tell you what, you've been working hard, i'll help you with this last bit.
Criminals don't own datacenters running the models that actually pose a threat.
Mid-high gaming pc money will have you running near-frontier level models locally.
train tho?
yeah it's always "we can run them on our gpus!"
yeah uh who trained it? how many gigawatts and thousand gallons of water did it take for inference?
near frontier levels of stolen shit, which is fine, it's all stolen, but holy hell don't act like it didn't require idiotic amounts of precompiling and a moral ambivalence to the stolen corpus.
Or maybe they're too stupid to know lol
No does anybody else, but apparently they have access to them well we wouldn't even be talking about it.
Of course every criminal has easy access to large data centers, and stuff. Except for the federal three letter criminals, of course.
Mathematics is a tool used to compute missile trajectories and weapon yields. Math should be shut down and deleted now.
LOL
Guns are a tool used to kill people, guns should be shut down and deleted now.
Knives are a tool used to kill people, knives should be shut down and deleted now.
Uh oh, you can’t point out mainstream cognitive dissonance you’ll get a down vote spanking.
That was my first thought. I think the difference is that AI isn't a literal killing machine. AI is more like a hammer, it can be used that way. A gun is a killing machine.
I think the difference is that AI isn't a literal killing machine
Tell that to the Iranian schoolchildren Claude thought were viable military targets (you can't, they're dead as hell), and to the people using AI drones to loiter and auto select targets in Ukraine.
A gun is a killing machine.
A gun is a machine designed to fire a projectile, what you aim it at is what decides if it's a "killing" machine or just a long distance hole puncher.
A bow is much the same, while it "is" a "killing machine," designed for use in war and hunting, the archers shooting at targets at the range near my house aren't murderers, they're just target shooters. Would they use their bow to defend their house, just as I may use my gun that has only shot paper to do the same? Yes, they might, but only one of us is apparently deserving of ire from the uninformed, despite being the exact same situation.
Technically, math helps us figure out the diameter and barrel length as well as the amount of accelerant to create adequate pressure to fire a bullet to accurately kill things, so in the end it's all math's fault.
NO use of AI is good.
I am engaging with this, because I want to believe Lemmy can be better than Reddit.
That being said, while I agree that LLMs and diffusion models should never touch anything close to art, anything that is life or death and anything legal or financial, it's important to remember, AI is an entire subfield of research in Computer Science, its many decades old and encompasses so many technologies that I could not even name them all.
LLMs and diffusion models are two of its many impressive achievements, we literally solved the natural language processing problem. It took us like 70 years, and because some silicon valley ghouls' rhetoric, we get (supposedly) well meaning people talking like Luddites.
Fundamentally correct, there is a machine now that can replace some workers, but technology isn't the problem, the capitalists are. This technology can make so many people's lives easier, so many things better and it doesn't need to be expensive or cause harm to the planet, because all of that is a fundamental problem with capitalism, this is not leaded gasoline.
NO use of AI is good. Period.
Whatever excuses you have is just going to sound like someone saying;
“Sure, ballistic misiles are bad, but fire is cleansing!”
all 98 comments