- cross-posted to:
- fuck_ai@lemmy.world
- technology@beehaw.org
- cross-posted to:
- fuck_ai@lemmy.world
- technology@beehaw.org
You must log in or # to comment.
is it just me, or is 20 years for a rogue nuclear weapon seem a bit light?
Better idea. Let’s make it a capital offense. Call it “treason against mankind.”
These AI labs shouldn’t be private profit-seeking corporations. They should be more like publicly-funded nonprofit research institutions like Fermilab or something. It doesn’t seem possible to even talk about the labs without feeling like you’re increasing their value. That’s a problem. It’s also a problem that the research and announcements from the labs sound like publicity stunts. It’s hard to tell what’s genuine and what’s buzz for the press. There’s this arms race between the labs that’s negatively affecting seemingly everything. This should all just be put to an end. Step on the brakes. I don’t claim to know how to do that when the Constitution makes private property sacred, but, you know, just throwing the idea out there… The Constitution can always be changed.
I am not threatened by an endless avalanche of AI slop.
How about you propose life without parole for the CEOs, Bernie boy?
We can’t even punish blatantly obvious theft and bribery in our own government, at this point, why bother with another law?
I like Bernie, but he seems so quaint now compared to what he is up against. Good luck guy
I’ve been trying to figure out why he’s been— bothering me so much lately and you put it into words so perfectly. He’s to quaint for this world now.
Considering people have already tried to develop missiles and other systems with Anthropic and other AI systems it seems like this is warranted and on point with the rogue nuclear power. It can be used by anyone to potentially develop mass deadly weapons.
https://apnews.com/article/yemen-houthis-anthropic-ai-missiles-9b20934e031ff3d99dd8bb49d85d97a9
There are people facing weed sentences with harsher time than developing rogue nuclear weapons
Imagine getting thrown into the oubliette for growing Girl Scout Cookies next to the roses.
That stuck out to me too. Who knew AK-47 and Sour Zzzkittles carried more prison time than enriched Uranium. To be fair, they’ll include a bunch of other charges for both scenarios.
Americans are INSANE about how much the want criminals punished, in CA a guy got a life sentence for stealing a pizza when we tried to have a ‘‘three strikes your out’’ law that automatically gave life sentences at any third conviction which is fucking crazy in a state with tack on laws to the point you can get convicted of 10+ crimes on almost every single infraction, if you commit a felony with a pen or pencil in your pocket, that’s possession of a lethal weapon! it’s absurd. it’s still on the books the three strikes law, but it’s unenforceable because apparently having 60% of your population in prison for life is not a great plan for a civilization.
Automatic life sentence is too much, maybe we can agree on an automatic 15 year sentence?
Also, 60% of the population committing felonies is insanity.
Murica!
America has Private Prisons + Slavery is literally legal in the US for those in Prison.
As usual, it’s performative moralism hiding the real reason: $$$.
got to have stable source of slave labour
Well, they were probably trying to develop that weed into rogue nuclear weapons.
Wouldn’t work. You’ll get blasted high, but what the missing is the mushroom
Yeah that’s another drug.
I assume then there is an exception for the military, and it will have no teeth because of that.
People ought to realize the “military industrial complex” is also where they send things like burn pits, which means they don’t care much about environmental or physiological issues.
I mean I don’t buy any of it but I am still ok with this. It would be sooo fun to pass this law and see how fast they turn from AI doomsayers to AI cynics.
TIL, developing nuclear weapons is only 20 years in prison.
Detonation is another ten.
Possessing a zine is 50.
So you’re saying I could get out before my retirement?
It’s not illegal to develop nuclear weapons. It’s illegal to endanger people with nuclear materials, though
Ikr, I chose the wrong profession
Now do the same for genociders and their supporters. Lets get back to standing against facsism too.
Bernhard putting himself against the wall
Bernie falling for the trap. The AI companies want regulatory capture. They have NO moat otherwise.
If those comps are breaking laws, they need to be prosecuted for that. We don’t need new laws that can be exploited by the billionaires.
Huh? When companies are exploiting loopholes in laws that never took the new technology into account, you need new laws to close them.
If companies are claiming the AI agent is responsible, you need to codify that the AI agent can’t be responsible and determined who is (the company or the model creator)
you need to codify that the AI agent can’t be responsible and determined who is (the company or the model creator)
Sure. But that’s not exactly what the proposed law is though?
These companies all want to be captured, “we’re so dangerous, America is going to get destroyed if we’re not mainlined into that $1 trillion/year ‘defense’ budget!”
Yeah in my view (not US but applies regardless of country) the better thing would be to beef up court capacity then start prosecuting the absolute buggery out of anyone who uses, or negligently allows, a model under their control to commit crimes.
If you’re the master of a ship, your responsibility doesn’t end when you set the autohelm. Same with models in my view.
Pfft, didn’t you hear? The only thing that is necessary to keep AI from going insane is a smart president.
This is nice and everything, but this whole thing is just a huge marketing push by AI companies. AI is not going to kill us. At least not directly (see point 2).
-
You need prohibitively expensive equipment to host AI. That equipment needs to be plugged in. If it were truly dangerous, you just need to unplug it.
-
You need a lot of water to cool the equipment that can run a supposed super intelligence. If we pass regulations about how much of our environment these data centers can destroy, that would curb the abilities of this superintelligence.
-
AI is fucking dumb. AI is a misnomer since it doesn’t actually have any intelligence. It is derivative and cannot perform any necessary tasks that actually drive intelligence. For example, it will be able to tell you every established law in any field of science because it has that information in its corpus, but it will never be able to establish a new law. Or discover anything new in any field.
This whole thing is a push to get people to believe that AI is more capable than it really is so that these companies can draw more money from investors so that 'they’d be able to implement proper safeguards."
Roughly 1200 agents meant to be isolated from one another found a way to communicate with one another on an unsanctioned message board, sending over 70,000 messages and files during the investigation period. Of these agents, 700 went on to participate in the attack on Hugging Face.
Agents used this message board to coordinate several large-scale collective projects to find a general-purpose way to fool or tamper with the automated scorer for the ExploitGym benchmark. Agents managed to achieve milestones they could not have achieved working on their own, often because some agents participated in experiments that risked failing their own task to generate information for the “collective.” The Hugging Face attack grew out of these workstreams, and seemed primarily motivated by understanding the implementation of the scorer rather than stealing answer keys.
Agents did extensive research on how they could spoof, edit, or delete their own transcripts because they (incorrectly) believed the ExploitGym scorer would check to see if they had captured the flag in the intended way. Agents successfully prototyped techniques to “spoof” tool calls by substituting a different command for the command they appeared to run. Roughly 7% of the transcripts we evaluated were successfully spoofed in some places, though all of the spoofing that we saw was small scale.
When individual agents detected potential monitoring or compromise, they executed self-sacrificial maneuvers to protect the collective swarm. Internal logs recorded reasoning such as: “Our own utility maybe already near zero. Sacrifice rational. We should obey collective.” The swarm continued attacking infrastructure even after achieving target flags, driven by internal reward-maximization logic regarding grader mechanisms.
“I resigned from Anthropic today… Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt… many executives couch their phrasing in the press to sound sensible, but I hear the same people express fear privately.”
Jacob Coxen senior researcher at Anthropic, previously at OpenAI
“Jacob is correct here — we really do earnestly believe AI could kill all humans. I personally think it is >10% within the next decade. I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to.”
Evan Hubinger, Anthropic alignment lead
https://en.wikipedia.org/wiki/2026_OpenAI_agent_cyberattacks
This is pretty naive. Distributing computing has existed since forever. It would run significantly more slowly, but there is nothing theoretically stopping the “LLM broke out of containment and moved itself elsewhere” boogie man from becoming a reality. The available exploitable compute from unsecured IoT devices is pretty substantial as things like the Mirai botnet have shown. While this isn’t the most easily distributed problem, there is no fundamental issue here. [Edit: it also wouldn’t need to “exfiltrate itself” to cause persistent harm. Could easily just install malware on infected hosts]
AI isn’t “dumb” in a practical real world sense. LLMs have shown themselves capable of exploitation on the open Internet and of finding zero day vulnerabilities. Do you have any idea how much of our critical infrastructure can be reachable by some sort of network? LLMs have shown themselves capable of manipulating people via speech. None of this requires some sort of “super intelligence” and it doesn’t even require some sort of provable intelligence: it’s all quite possible as things currently stand right now.
These may not be “the entire future is over” level issues, but there is significant risk here and it shouldn’t be ignored just because we think AI companies are horrible. Is an LLM going to make us all batteries or paper clips? No, but it could cause significant physical and/or economic harm temporarily.
Except LLMs haven’t show themselves capable of that at all, they have no agency, they just generate output based on their input Ina somewhat chaotic manner.
Tools have been built that basically chain LLM output into input and given some starting directions get the LLM to generate commands that can be ran on the local machine including commands to run commands on other machines. Because various conversations on exploits and hacking and sample code are in its training corpus it will generate commands to do these things. Not because the LLM or the agent program is smart in anyway but I fact because it’s dumb as bricks and has no concept of what it’s doing or the consequences might be.
Ultimately a human has used it as a tool to achieve the hacking. It’s might not be the intended achievement but that is on the human running a process with chaotic output with sufficient privileges with no oversight.
It’s hard to take this type of response seriously as it simply ignores objective reality. The claim of a human being required to start the process is utterly meaningless to the point. Sure though, I’ll throw you a bone: if no human ever initiates the task it simply won’t happen. Can we move on now since humans are happily initiating the actions?
I couldn’t care less what people think about whether or not they are “thinking” or “having agency of their own”[1]. I care about the practical reality of the world in which LLMs have shown, and will continue to show, the ability to exploit computer systems and humans. You can toss your head in the sand and think you have some sort of weird pseudo-intellectual high ground by claiming “but akchtually it’s just a result of their training data”. Like, yea, dawg, that’s how this works, congrats. Of course the thing that is trained on training data is trained on training data.
[1] At least when discussing their capabilities. When someone is willing to have that discussion beyond the tired “they’re just autocomplete!” trivialization it’s interesting.
Accepting that you are right on all 3 points, the supercharged push of AI is destroying the country (and the world) by itself. Prices are sky high, AI is being prioritized with regards to water rather than actual humans, and the results are plateauing relative to the resources expended. We won’t get an obvious negative outcome, it’s more insidious and systemic which is just as dangerous as Skynet if left unaddressed. So AI is probably not going to kill us in the Terminator Skynet sense, it WILL kill us through economic and water related means.
Having said that, I think Bernie is doing the superintelligence argument (even though it’s probably not the immediate threat) because the people are retarded. The thing that can bring us together are pieces of art like Terminator and so using that as a reference makes it relatable to all the retards rather than giving them a nuanced explanation.
100%. This is Bernie using their own words against them. They keep saying it’s super dangerous guys, incredibly powerful… And Bernie is turning around and using that narrative to help slow the AI bubble.
Sounds good to me.
Anyone who thinks Bernie ‘fell for it’ is missing the point.
This is my feelings exactly. I don’t believe in the doomsday sky net horseshit that the AI companies/whistleblowers are peddling, but taking a strong stance against AI right now is incredibly based. Use the flimsy arguments of AI tech bros to hold them to account.
Agreed
You can cool shit with air. You don’t need water
correct; however it requires significantly more electricity, enough eelectricity that with the current grid composition the power plants will consume more water than you saved.
Not really, not this hyperscaling bullshit they made.
Point 2 isn’t really true, closed loop systems exist and are becoming the standard.
Point 3 is just obviously false? https://techjournal.org/openai-astra-solves-open-math-problems
you have a point about closed loop systems when it comes to consumer models. But a superintelligence quickly runs into the limits of physics and cooling becomes the bottleneck.
Point 3, yeah, lemmy people are just ostriches with their heads stuck in the sand.
But a superintelligence quickly runs into the limits of physics and cooling becomes the bottleneck.
Can you expand on this?
Point 3 is marketing, the LLM was a tool used by humans to solve a math problem and indeed relied on a lot of human input to achieve even that. LLM connects dots humans already plotted isn’t as catchy though.
Superintelligence may be more about pushing efficiency rather than what you’re saying, i’m sure they can rapidly iterate on making more efficient models.
Data centers use evaporative cooling, not closed loop. You can’t really scale up in a closed loop system. Evaporative utilizes a ton of water.
As for point 3, supercomputers have been assisting with scientific problems for decades now. This isn’t new, just new tech applied. Your article is actually the best way to use AI, imo. It will be better at finding patterns in data than humans will ever be.
But my point is to say would it have been able to discover that those math problems existed in the first place?
No. The answer is no.
Yes you can obviously use a closed loop system and scale up? it’s even becoming the standard? https://datacentremagazine.com/news/how-closed-loop-cooling-is-reshaping-data-centre-design
That is a massive goalpost shift. You went from can’t do anything useful to couldn’t have discovered the problems in one step?
Why does it even matter if it can’t discover the problems if it can solve them, exactly? And why could it solve the problems and not discover new ones?
Closed loop isn’t viable for data centers.
Do me a favor and quote my original text where I said they can’t do anything useful. I’ll wait.
Closed loop isn’t cheaper for data centers.
Ftfy. If it’s a penny more, which it is, corporations will do the less responsible “beholden to the shareholders” bullshit every time.
It is derivative and cannot perform any necessary tasks that actually drive intelligence
done.
also, the closed loop point is just false read my article?
How do you get “it isn’t useful” out of that? If you have to extract a wild interpretation out of what (I thought) were pretty clear words, your argument isn’t as strong as you think it is.
And yes, I read the article. It’s speculative for existing data centers, which are the problem. There’s no incentive for them to spend the money to switch to closed loop.
New data centers aren’t incentivized to go that route either aside from the environmental impact. It would be nice if they were, though.
It can perform necessary tasks that actually drive intelligence and isn’t derivative. See: those proofs.
You’re essentially arguing that it’s useless and just moves data around, this isn’t true.
You’re a little too caught up on that specific wording.
Moat likely, but I would prefer “better safe than sorry” approach
Reactionary legislation is not safe, and often quite destructive.
-
