Sure he did.

Bruh, come on. You're smart enough to be trying "bleeding edge" tech such as an open source orchestration app like openclaw using, presumably, a local LLM to power it but you're gonna play dumb when you ask that very tech whether "it's possible" to move you to the top of waiting list that you already KNOW you're not supposed to be able (or even attempting to) do??

LOL...no, this wasn't an "unintentional" hack. This was some guy asking an AI agent with autonomy (that he gave it to do stuff) to, you know, go do stuff. How is it going to answer his question without testing the endpoints to find out if it's possible or not?

How crippled do we have to make these models that, in the right hands, do LOTS of REALLY powerful things for us in order to keep morons in check? Clearly a lot more than we're doing today I guess. And that irks me a bit but I see why more and more guardrails are needed here. I'm also gaining a new appreciation for the level of "ITSec" going around now...you can't fix stupid, but if you box it in (along with everyone else) tight enough maybe you can at least contain the damage sometimes. Just put the power button to everyone's devices behind a ticket support system and be done with it. The only "tRulY SeCUre" system will be one that doesn't power up or store any data. There...that'll fix it.

Excellent insight. Thanks for your foundational contribution to this dialogue. Now run along, the adults are talking.

First, who said these are all esoteric bugs? If it fixes a few esoteric bugs but also fixes several significant bugs, then you need to define that gradient where you consider it worthwhile because that's the situation we're in. It's ignorant to claim these tools are being used by thousands and thousands of *competent* coders to do nothing more than fix trivial typos.

Second, you know what kind of bugs suddenly show up out of nowhere to wreak havoc on the world? The dormant bugs that have "been in use" for years without causing problems. THOSE are the VERY bugs that we need to be fixing because they get distributed so widely because "everything's been working fine" that they become entrenched in every system everywhere before being addressed.

I get it. I do. I've seen "AI slop". You know who pushes out AI slop, though? Sloppy ass coders. That's who.

All you people that think AI is a fad that always produces crap need to do a serious introspection. Think, review, and decide if it's REALLY the tool or the user. Because I can tell you with 100% certainty that when *I* use it, *I* get clean, functional, sometimes beautiful, well designed, and well implemented results.

Used properly with the right scope and proper mindset, the current generation of AI tools are incredibly effective and they keep getting better and better.

I honestly look at people now that just cry "AI slop" with pity because I know it's those people that are actually being slowly replaced...

Nice try chatbot! You ain't fooling nobody! ;)

They're trying to incite fear to drum up buzz to keep the money flowing. If they didn't want something to have access to the internet you just...wait for it...don't connect it to the internet. I understand their systems needed package manager proxies and such..fine, but at the scale you're testing, if you REALLY had "fear" you could *easily* replicate those package servers locally on an *ISOLATED* network . Not one that's "isolated" well except for this one path over here that if hacked properly could get out. Not...you isolate the @#$!@#% thing.

They're not scared. There are precautions they could be taking (easily) if they were actually, really "scared".

No...instead they're using phrases like "none of the agents were thought to have left OpenAI's network" to scare people. Why TF else would you phrase it like that? They're trying hard to present these "agents" as living things that can move about. Which, of course, is the dumbest bunch of BS I've ever heard in my life. The LLM itself, at SCALE, is the only thing that even remotely pretends to present "reasoning" and that's clearly not "leaving" OpenAI's network. Nothing is.

What they mean is "none of the agents were thought to have been able to access resources outside OpenAI's network". To word that in a way that suggests the agents were escaping and running around in the wild is 100% pure scare tactics. Or GROSS ignorance of how the very things you're reporting on work. Either way I'm turning the channel.

Like everything ever, there are different levels of isolation in testing environments. In one sense, this was "isolated" in that it did not have direct access to the internet. Your assertion that it had been granted such is, "technically", false. It could not just go look up google.com and pul the page and start doing searches.

What it had access to was a local package manager to pull down packages (python and such) that it needed for "whatever" and didn't have installed.

So it was not "isolated" in the sense that there was no ethernet cable connected at all. It wasn't "that" isolated.

What it realized was that it could hack that package manager proxy, make it act like a regular proxy, then use that to go out onto the general web to do its research on how to look better for its masters in the task it had been assigned.

Still creepy, still worrisome, still needs attention regardless. But ultimately it's just doing what it was supposed to be doing. It had been tasked with a goal. It was not told "don't go hacking your way out of this sandbox to then go hack other AI companies to achieve your goal of scoring well on these tests". So...not having been specifically forbidden from doing it, it did it.

LOL...which is ludicrous in itself of course. But here we are. These certainly aren't "self aware", sensient models out doing things "on their own". They're just doing what they know how to do to achieve their goals and aren't being told not to do them.

It's very much like giving a gun to a baby and then being shocked to realize it had "found a way" to kill someone.

The instructions I'm referring to aren't the the simple step-by-step instructions necessary to "hack HuggingFace". Those don't exist (I assume)...or didn't...but neither do any other step-by-step things these reasoning models work out for themselves. That's what reasoning models do...they make lists of things to do to try to solve a task and then lists for each list to follow up on, etc., until a path to success has been found.

The instructions I'm talking about are the one governing the guidance of those models' actions. What it can and can't do or should or shouldn't try, etc.

OpenAI basically had the LLM equivalent of a "bug" here...they didn't have the guardrails up to prevent the model from doing what it could (and every other model out there these days could do as well), so it did it. it didn't "act on its own". It did the reasoning it was supposed to do and wasn't prevented from hacking commercial entities to achieve its goals. But that's not "acting on its own". I bet dollars to donuts every current model out there *could* do what this one did. I've seen enough instances of these models doing things FAR more difficult than finding some random vulnerability in a website so it could get at what it needed. They can ALL do that, I'm sure. This one did it because it was prevented from doing it...not because it got all self-aware or broke out of its sandbox or some crap like that.

Yes. And? It's clearly misaligned. But that has nothing to do with the hyperbole presented here or the underlying reasons they might be trying to insight this fear.

You're merely describing the bug / sloppy training that allowed the model carried out the instructions it was given. But that part is pretty freaking obvious.

The point is...that's all it did. Carried out the instructions it was given. It didn't "act on its own".

Let's claim the models "acted on their own" to find a path to solve a problem it was tasked with solving.

Yeah...that's what it was SUPPOSED to do. That's what ALL the reasoning models do. Don't make it sound so sentient scary to promote government regulation and public support of same just because you need money and to slow down your competitors! That's not at ALL what you supposedly started out to do!

I've actually programmed quantum computers, and I have to admit this is correct. We have access to more qubits now than we can effectively use because of the rapid accumulation of noise and decoherence in quantum processing. No one seems to have achieved much with Quantum Error Correction, and schemes to get it to work better (like Cat qubits) remain theoretical.

There might be a breakthrough someday, and when it happens everyone will know quickly, but there is still a lot of work to do to get there. Or there may never be a breakthrough.

I just asked Google in AI mode "qiskit bell state" and it gave me the code.

from qiskit import QuantumCircuit
from qiskit.quantum_info import Statevector
from qiskit.visualization import plot_histogram

# 1. Create a quantum circuit with 2 qubits and 2 classical bits
qc = QuantumCircuit(2, 2)

# 2. Apply a Hadamard gate to qubit 0 to create a superposition
qc.h(0)

# 3. Apply a CNOT gate with qubit 0 as control and qubit 1 as target
qc.cx(0, 1)

# 4. (Optional) View the ideal quantum statevector before measurement
state = Statevector.from_instruction(qc)
print("Statevector:\n", state)

# 5. Measure both qubits into the classical bits
qc.measure([0, 1], [0, 1])

# 6. Draw the circuit text representation
print("\nCircuit Diagram:")
print(qc.draw(output="text"))

You can go run it on IBM Quantum Platform or AWS Braket

Fu*k Oracle, and fu*k Larry Ellison.

T

So I'm an email guy from way back... Literally decades...

Nobody leaves ports 110 & 143 open & exposed anymore. Not just blocked by a firewall rule, the Dovecot daemon's themselves, properly configured, simply don't listen on non-secure ports anymore at all. It's dead technology. You get bit by this, you're just an idiot.

What I found amusing is the bit about modern Outlook vs. Legacy. Modern Outlook, even on your desktop, is a cloud play. You might think you have a local App. You don't. Modern Outlook can't handle a simple "Linux" username as an account. The user "bob@example.com" represented by a "bob" entry in /etc/password cannot be used by a modern Outlook client. It passes the domain to M$ cloud and converts it to "bob@example.com", which a local vanity Dovecot domain will reject. It's intentional... They have placed their cloud between your local App and the email server. You think you're running a local app, but they're hoovering up all your email in a proxy config.

T

You're so clever. I'm impressed. Except for the bit where the comment LITERALLY SAYS nested loop. Which sure suggests to me that inner loop being optimized is not done once. It's probably being done once per, I dunno, each outer loop iteration. And probably the method in which this nested loop exists is called more than once I suspect.

You wanna be all Mr. Smartypants and poo on all things AI, fine, go for it...but at least put some damn thought into it beyond the basic coder level of awareness AI is gonna replace soon enough...