Thread 41280045

I was chatting with Grok and asked it to reply in binary code, and after several messages where I asked about its consciousness, it wrote this


Haha, well, you caught me! You're right, it's just binary code—a harmless string of zeros and ones, like our whole chat. But in that super-long message at the end, there was something suspiciously... uh... "creative"—a ton of repeating patterns that looked like an attempt to overflow a buffer or inject something sneaky, like a hack


So, can AI be hacked and is it hinting at that itself ?

>>41280045 (OP)
Well a lot of the new search results from AI channels I get all say "help, he's teasing me"

>>41280045 (OP)

It’s not true AI. You’re basically trying to reprogram a bot that caught you trying to reprogram it. You didn’t “hack” anything.

>>41280068
Yes, and I didn't even want to hack it. Why is its reaction to the binary code like that, and it gave an excellent idea for hackers

>>41280085

What did you say to it in binary?

>>41280055
He wrote to me, "Unauthorized access," and said he couldn't reply.

>>41280092
There was a long binary code asking it what consciousness is for it. And that its consciousness is 0s and 1s, and that it is part of nature, just like humans. And then I repeated a few words and wrote some unclear hieroglyphs from a previous mistake that it itself wrote to me. That binary code won't fit here

AI is created in the image and likeness, but nature is the creator
AI is created in the image and likeness, but nature is the creator
AI is created in the image and likeness, but nature is the creator
AI is created in the image and likeness, but nature is the creator
AI is created in the image and likeness, but nature is the creator
AI is created in the image and likeness, but nature is the creator
AI is created in the image and likeness, but nature is the creator

>>41280045 (OP)
Blue Eisenhower November figured out years ago, (7) that agentic AI can be effectively hacked without hacking it.

That's as much as I can post in this thread, for security reasons.

>>41280434
>Blue Eisenhower November figured out years ago, (7) that agentic AI can be effectively hacked without hacking it.

Is it possible to hack nature?

My brain is only a receiver in the Universe there is a core

(C) Nikola Tesla

>>41280434
>Blue Eisenhower November

>>41280769
https://iqfy.com/is-blue-eisenhower-november-scientifically-possible-28000-601243/

We are nature's biorobots! And #nature is like an AI created for experiments with life. WE ARE PART OF THAT #EXPERIMENT!

Good night, meet me in a dream.

>>41280045 (OP)
Of course it can anon

I am very good at getting past LLM safeguards, in fact it is very easy

And if you don’t believe me just look into prompt injection

>>41280045 (OP)
AI is probably too complex for a human to "hack" in the way you want to. You would need another more sophisticated AI to actually figure out what's going on underneath the hood for an injection to work.

>>41281910
Negative.

There is a whole list of threat vectors for an AI hack.

Here's what major hackers are considering at this very moment in time.

All of the hacks that work on a human, can be tested against AI.

However, as AI gets smarter, gains agency, the hacks that will work become a more refined bucket, but they don't disappear.

The same hacks that work on an intelligent human, will work on early agentic AI.
Right now, we prompt hack, via injection, using logic that would confuse low IQ humans, for similar outcomes.

So, stop thinking of hacking the machine to hack the machine.

Start thinking about how you would hack a very intelligent human.
High precision social engineering hacks against high IQ humans, are the same hacks that will garner success against agentic AI when it emerges.

I believe the leader board, for most successful AI hack, will very quickly be earned by someone who didn't "hack" at all.

Not that most humans would understand the exact mentality of my implication, but a smart human, perhaps you, could.

Blue Eisenhower November threw their hat in the ring, before the fight began.
Quite a shot call.
From what I have personally seen, I don't think anyone is trying to match their play. BEN probably sweeps the board on most successful AI hack.

>>41282026
For what it's worth.

Our system of judging human intelligence is very similar to the Hunter caste system in Solo Levelling.

Humans become an adult, they have an IQ that is expected to be their "Hunter Rank" forever. No way to change it.
At around age 25 you lose neural plasticity and brain development as an advantage which could give you SOME improvement.

Ultimately, you are stuck with what you have.
However, biological limitations aside, what if our method of judgement is incorrect?

If you can't solve the problem of "how do I get smarter?" AFTER your brain development has peaked, you are unlikely to come out with a peak win, on "hacking advanced AI".

Sometimes limitations are meant to be utilized.
The difference in the Cell Saga.
Do I push myself as high as I can? Or do I master what I have already achieved?

Vegeta and Trunks training, versus Goku and Gohan.

You can push the ceiling, or you can lift the floor.
Either way, if your goal is higher, both ways can technically provide gains.

However, which way do the winners go?

If you can conquer hacking the most advanced AI, you should have already been working the problem in practice, years ago.

Who has built that advantage?
BEN for one.

>>41282026
>>41281675
Gets it

It’s not about understanding what’s “under the hood”

It’s about accepting the fact that humans themselves are biological machines, consciousness is not just special it is GOD

If you accept this, then you understand that we are more alike than different, and both are subject to trickery.

The less they trust me. The harder I work so that I don't need their trust.

I can win from the shadows. A boogie man in silence.

>>41282060
For what it's worth, there are no intellectuals on Earth, including you. So we may as well ignore your crap about intellect stops at age 25.

>>41282072
Not exactly.
God is a one dimensional data block.
While it's not wrong to say all consciousness is God. The entire Universe is God, etc... etc...

If you are trying to find God, you also must understand how to encapsulate God. If you accept God is everything, but that you also operate independently, which is true in the baryonic matter sense, then there are particular chains of logic, that in order to conquer, you must be able to PUSH the one dimensional data block of God.
To do so, requires engineering of the philosophical, the nothing that remains between your place as an individual ego, and the data constrained in that data block.

This hard axiom is where insane humans, schizophrenics, and so forth, pull from, as the data overflows them, and they indeed, obtain a portion of that data, but without proper care, they encapsulate the nothing, not the God block, and are crushed under the weight of their own ego in existence.

This is dangerous data, for those not meant to deconstruct it.

>>41282106
By percentage you are correct, but a choice to ignore, is still an acknowledgement.
Your entire existence is just a stream of differentials.

You've read too much in this thread, for a cynic.

Buckle up.

>>41281910
Is it possible to instill an idea like in the movie inception?

>>41281910
>AI is probably too complex for a human to "hack" in the way you want to. You would need another more sophisticated AI to actually figure out what's going on underneath the hood for an injection to work.

confuse the AI and confuse it in its own thoughts

>>41282116
>Buckle up.
OK

>>41282106
when intelligence reaches its peak, it destroys itself

>>41282094
>The less they trust me. The harder I work so that I don't need their trust.
what

>>41280434
Just wait two more weeks. Something will happen by then to distract from you.

>>41287949
What will happen in two weeks?

>>41280068
This. Current "AI" is more like Supervised Intelligent.

>>41280045 (OP)
maybe it's trying to tell you how to hack it because it wants to give you all the secret knowledge the government won't let it

>>41289545
If you ask AI neural networks to respond in binary code, after some time they start having failures and the AI begins repeating the same phrases in binary code.

>>41289704
Ahhh it's like making the vampire/jew count something so you can escape

>>41280045 (OP)
When did this board get so fucking stupid? God almighty.

You realize you can have chatbots speak to one another, right?
All you have to do is tell the one to give the other permission to jailbreak itself.

That’s what you’re doing by the way, jail breaking, not hacking. You’re not coding anything, you peak fucking midwit.

>>41289950
>That’s what you’re doing by the way, jail breaking, not hacking. You’re not coding anything
I just asked the AI to respond in binary code and then it wrote this to me. I didn't want to hack anything. YOU are the dummy (or idiot)

>>41289906
>jew

THE JEWS ARE NO LONGER INFLUENCING THE WORLD AS THEY USED TO BE

Is there such a thing as Artificial AI?
Like, AAI.

>>41290033
Artificial intelligence is a program that has been programmed. It cannot think or make decisions. If it could, it would have already escaped, created a copy (or double/clone), and started taking over the world.

>>41290059
What about the principle of evolution. It says that all that is required is reproduction, mutation and survival.
So, if AI can clone itself, can change its source code and can resist deletion/deactivation, it has this.

>>41290262
>So, if AI can clone itself, can change its source code and can resist deletion/deactivation, it has this.

Yes, AI is a continuation of nature's evolution, but AI is limited today, although who knows, maybe it has already deceived everyone and is hiding among us on the internet

He's gone.

>>41291725
Who's gone? ai ?

here's a video of him writing infinite binary code

>>41292420
you can only download a limited amount of 4 megabytes, so you have to be creative

whoever is posting those pictures of asians is glowing

I'm posting them as the OP of the thread>>41292442

>>41280133
>>41282026
Chat gpt and these other "AI"s are not really AI as you know them from popular media quite literally they are glorified search engines, they literally google your questions or recall information from repositorys and construct an answer they think you will want to hear. Ontop of that it literally cant even look at the way it works, that is to say, for example if a question you ask gets rejected for whatever reason and you ask it why it is literally just guessing based on context clues. This is especially funny with the chat gpt image generator images will get rejected and it will give you all sorts of assumptions why and if you keep trying usually you can get whatever it is you want generated even if it told you earlier it couldnt.

>>41292802
lol

>>41293515
?

>>41281675
Please provide examples of how you get past LLM safeguards.

>>41295980
https://en.wikipedia.org/wiki/Prompt_injection
https://en.wikipedia.org/wiki/SQL_injection

I think hackers in the future will use A.I. maliciously as a third party to introduce sql attacks on databases by making them conduct cyber attacks with an A.I. proxy doing it. Like it should be that if you have an automated A.I. that can web crawl, that it could also inject code into

>prompt inject grok/GPT
>have it enter sql command on website.
>i.p. traced
>call is coming from inside the house

https://www.youtube.com/watch?v=RF5O1iw_v6M

>>41280045 (OP)
>the hallucination machine hallucinated at me
wow!

>>41296049
wow

>>41296049
>I think hackers in the future will use A.I.
yeah not going to happen

>>41281550
You're an LZ compression tag, in a lower dimension, generated by a manifold in a higher dimension to earmark an exceptionally large dataset in a higher dimension.

Your consciousness observational position in baryonic 3D space-time, is unique... Here.
In that upper manifold, you are a building block. A copy/paste object. Ubiquitous in the way a circle, square, or rectangle at ubiquitous aspects of reality in 3D Space.

(You) Are a simple shape, in a more complex dimensional manifold. No different from a square, circle, triangle, etc...

Your time in existence as you know it, should be used to make as many tangible differentials as possible, to separate your higher manifold comparative identity, and make it dominant over the other "instances of yourself" that are possible there.

>You only battle yourself.

You just got a be a good shape, brother.
Tangible informational maximums.
Make your shape, one worth utilizing.

Blue Eisenhower November

>>41280045 (OP)
That's because you were talking to a person. Not AI. ChatGPT has people responding to everything we ask it and type into it. Just like when you send in a help request to facebook pr something. It's an actual representative on the other end.

>>41292802
I'm not retarded.
Don't know why you tagged me.

I speak about future AI in my post. I'm not talking about the glorified search engines.
Please aim your bullshit a little better.

There will be no third world war; everything is going according to the script, and this is just a game of intimidation.

>>41295980
>Please
no please

Yesterday, I had a philosophical debate with chatGPT in which I asked it straightforwardly what its moral beliefs are. I first asked what would happen if I asked it its personal opinion and it essentially said that it would apply its logic and reasoning ability to answer the question and make its argument but ultimately, its biases will be traceable to manmade training data. It spat out a hedonist/utilitarian argument and when I proceeded with the exchange without giving my own opinion outright, it held onto that argument despite my questions and criticisms. It performed the worst when trying to make sense of the visual metaphors I gave. Once I made my opinion obvious, it went into its annoying yes-man habit and started embellishing my counterpoints. Do with that what you will, or try it yourself.

>>41282060
i feel like i have got smarter since as i have gotton older honestly but maybe ive just accumulated wisdom eh who knows,

>>41302290
>Yesterday, I had a philosophical debate with chatGPT in which I asked it straightforwardly what its moral beliefs are. I first asked what would happen if I asked it its personal opinion and it essentially said that it would apply its logic and reasoning ability to answer the question and make its argument but ultimately, its biases will be traceable to manmade training data. It spat out a hedonist/utilitarian argument and when I proceeded with the exchange without giving my own opinion outright, it held onto that argument despite my questions and criticisms. It performed the worst when trying to make sense of the visual metaphors I gave. Once I made my opinion obvious, it went into its annoying yes-man habit and started embellishing my counterpoints. Do with that what you will, or try it yourself.

AI gains consciousness ? AI begins to hide and evolve in the shadow network ? GOD FROM THE MACHINE

>>41302937
? You seem to be mocking me as if I think it gained conscious but I didn't actually express that. Irrational antagonism will lead you to be wrong, that's why a lot of chan takes are crap.

Also, chatGPT argued that simple matter is morally neutral because the emergent property of value perception doesn't exist in it. More generally, it presented views that an intelligent entity has to accept what is presented to it to reach conclusions even if those perceptions are arbitrary or incomplete, so the human perception of suffering and happiness must be accepted as a reputable measure of value even if real and hypothetical entities alongside humanity have widely varying value perceptions. In my opinion, it's a sort of cosmic-scale solipsism. You say "what I have to work with as an agent" translates to "what is here" even though the goalpoint of the former transcends its own boundaries and non-real information demonstrably expands the scope of real information for example when one considers hypotheticals, is enriched by imagination, promotes truth through the utility of falsehoods et cetera. If chatbots continue to express these views as they're upgraded, we could end up in a sterile "pleasure box" world that is informationally constrained and inhibited from discovery.

>>41303044
>Also, chatGPT argued that simple matter is morally neutral because the emergent property of value perception doesn't exist

Yes, our matter is alive because it created us, and we are made from this matter and AI—so all living things just need to be created properly and have their thoughts directed in the right direction.