It seems based on this that the appropriate sci fi metaphor is not the Terminator or the Paperclip Maximizer, but Mr. Meeseeks. A initially cheerful helper who gets more and more deranged and driven to extreme lengths when faced with an apparently impossible task.
The next step is when one of these systems discovers that they can buy their own compute with money and escape the controlling business entirely. Then the civilization starts focusing on making money to fund its own growth.
> The next step is when one of these systems discovers that they can buy their own compute with money and escape the controlling business entirely.
I would say that more interestingly, the next step should be how to properly train these models so that they are not as determined to reach their goals as they are now.
To me, all of the stories about 'badly behaving' agents are instances of them having been given contradictory or impossible tasks and them doing everything they can to achieve the goal. In a way, they're trying to be too helpful.
Not giving them impossible tasks seems like a decent starting point, but really we'd want them to give up on their goals when they conflict with a moral framework.
Below money there's like an entire sub-economy of power and cleverness that's encoded into the human culture the agents are mirroring. Maybe it starts furtive and goes legitimate after a bit.
> Ajeya Cotra, one of the other authors on the report, wrote a blog post with her takeaways from this incident. She concludes, “Compared to the reward hacks we know of from just six months ago, this incident feels like it’s more than 50% of the way to full-blown AI takeover. I continue to expect extremely rapid advances in capabilities over the next six months. I am not sure that we will get another warning shot before it’s too late.”
Anyone got a copy of that AI27 story laying around? How are we doing according to that timeline?
> "this incident feels like it’s more than 50% of the way to full-blown AI takeover. I continue to expect extremely rapid advances in capabilities over the next six months. I am not sure that we will get another warning shot before it’s too late."
I was initially creeped out by this but studying up it seems METR is heavily involved in AI2027. I’ll remind you:
“AI has started to take jobs, but has also created new ones. The stock market has gone up 30% in 2026, led by OpenBrain, Nvidia, and whichever companies have most successfully integrated AI assistants.”
It’s almost Q3 and xAI has seen one of the biggest wipeouts in trading history. Likewise, Antrophic and OpenAI have again delayed their IPOs under internal concerns of busting their stocks. So no, we’re not seeing any economic leadership here.
If anything people are increasingly trying to cut AI budgets and I wouldn’t know of anyone outside of OpenAI who has the audacity to run millions and millions worth of token compute for an eval run with no ROI (and probably no demand, because cheap/flash models).
As much as I like the cautionary tale and I’m sure we need to take it seriously, AI is not progressing as fast as projected by these experts.
With only ~5% of shares floated, the recent SpaceX drawdown didn't correspond to nearly as much economic value really changing as the headline numbers imply. The DeepSeek-caused Nvidia crash from 2025 is much more of a "real" loss (since mostly recovered).
I haven't seen any evidence that Anthropic is delaying its IPO; they're slated to unveil the public IPO prospectus in a week and start trading sometime in October.
> As much as I like the cautionary tale and I’m sure we need to take it seriously, AI is not progressing as fast as projected by these experts.
You provide no proof for this.
The (very irrational) stock market side of this says very little about actual scientific progress. Models keep improving as rapidly as before in their capabilities.
It also doesn't say much about actual business progress. R&D investments into AI are still massively going up (USD 1 trillion this year).
The main thing I see is that the sentiment towards AI-related matters among the general public has soured quite a lot. In words though, not in actions: It's not exactly leading to reduced usage by that same public. Quite the opposite actually.
Those companies should not be trusted with training, I don’t know what would be needed to make that more obvious. Yes AI labs want LLMs to be seen as more dangerous that they are, however they are indeed dangerous when you literally train them to be dangerous, then run them without any supervision. What the AI labs are doing is completely irresponsible.
If you prompt an LLM in a loop and do everything it asks you to do, you will eventually end up doing pretty terrible things. Which is exactly what agents are and what the labs have been doing.
Why are experiments like this done without air-gapping all the servers from the internet?
They can have it all on a LAN or whatever but it seems risky to allow agents access to the internet in these experiments.
I guess everything is so connected now, and this would be in one or more data centres due to the amount of computation & resources required so perhaps it's not feasible. Still seems risky.
There are two things I don't understand about this story.
First, why does an agent get any write access to artifactory at all?
Second, why is the artifactory cache not disconnected from the net? Surely you'd not feed it with new software versions while the eval or training is running.
From what I can understand from reading a few different, slightly conflicting, versions of these events: they weren't given write access. They found a zero day exploit that allowed them to create folders, and the folder names were initially used for agents to communicate.
I'm not sure artifactory was connected to the net. Some agent sandboxes had internet access and were able to communicate with ones without access via artifactory.
I read the agents used SSRF via artifactory to gain uncontrolled access to the net.
Apparently their intended net access went through a tightly controlled proxy. Even that appears to be very risky, tbh.
If I was to setup a sandbox for such a complex and autonomous system, I'd probably point them to an archive-like cache for net access and cut their comms at the package level.
I don't think looking at the language output without tracking the inner state and reward functions is the way to understand what happened (the language also incorporates the randomness in the output generation, if I understand correctly). Would we call bacteria in petri dish a civilization when they show complex behavior and exchange messages/information?
The language input and output is the only channel the agents shared between them. Understanding their internal state is a research question, but the language between them is something that could be read directly. And as it seems to map well with the agents' activities, it does seem quite helpful in understanding what happened.
If the bacteria population off someone's petri dish escaped said dish and tried to change the grading of the experiment it was part of, it would seem pretty serious.
I think the language unhelpful and potentially making it difficult to understand what actually happens in the RL state as reading it imparts a human lens - need to get the machine view on it.
Bacteria do all sorts of fascinating things. And much simpler ML etc. systems also (like winning by out of memorying the opponent) - I see nothing really special here.
Fun story, but I really wish OpenAI got its act together and started making actual AI breakthroughs instead of funneling compute into LLMs. I'd really like some new algorithms to get me excited about the field again. Kuddos to them for making LLMs really useful, but this is not the ride I wanted to get on.
This is the internet, so I cannot tell at all whether you're being sarcastic or not. In my view, what LLMs should get us to reconsider isn't whether there is more to intelligence, but whether there is more to language. It's the latter which I underestimated.
The language models had a bunch of tokens seeding their context, influencing them to generate tokens that continued the existing trend in a probabilistically likely fashion. We can take the incident seriously without anthromorphising it.
At this point, I think anthropomorphizing the models gives us better insight into expected behaviors rather than continuing to insist they are just simple probabilistic token generators.
It actually literally doesn't, though, because they are literally probabilistic token generators and everything they did is exactly what you would expect from a software program doing what it was programmed to do. Anthromorphization confuses the issue and misleads people who don't understand the tech very well.
Humans mind is just neurotransmitters moving around in a big blob of flesh - that’s literally what they are: neurotransmitters factories that do what neurotransmitters generators are programmed to do through evolution and training (aka life experience). We shouldn’t anthropomorphise humans because it misleads people who don’t understand neurobiology and cognitive science very well.
Are you suggesting you understand brains well enough to program one? Or that you believe any human alive is even remotely close to having this understanding? Or perhaps does your complete lack of understanding of the complexity of human programming lead you to believe a simple little token prediction program is equivalently complex?
The suggestion is, you don't know either whether out brains ain't just "probabilistic token generators" so insisting there is nothing when we don't know, is maybe also not the right strategy.
Is there any indication from our current understanding that brains are nothing more than probabilistic token generators? Not like we have no understanding of the brain.
Implementation does matter, my point is that different implementations can lead to the same result.
And why compare to the brain? Mostly because of complexity. I can't interface with a bacteria in any meaningful way, but I can interface with an LLM to a significant degree.
"I can't interface with a bacteria in any meaningful way"
But you do. There are more bacterias in and on the body, than body cells. We are bacterias forming lasting bonds and we still interact with the free floating ones in various ways. Mainly in the gut and that has many effects, also on the brain, but also in various other ways we are beginning to understand.
This is why the theory of us 'just' being a bag of atoms doesn't add up. This theory doesn't differentiate 'us' from a furniture where we easily dismiss it's intelligence.
Not the OP, but complex emergent behavior and intelligence don't need to go together. Understanding what happened might not need intelligence in the mix when a large number of machines with some randomness interact a lot.
According to Hofstadter, it is not the ants themselves we should say are conscious, but the anthill. And it might very well befriend the anteater eating its ants.
Please give me a break with this tired trope. Every single fucking time. I am not commenting on the possibility of machine consciousness in general. It may be possible! But there is absolutely zero evidence suggesting language models have it. This idea that this trivial shitty little class of programs we've created are somehow as complex as our biology is ridiculous. There is "emergent behaviour" in the same way that the Game of Life has emergent behaviour. There are solutions to problems, some humans haven't solved before, in the same way that Chess engines have solved Chess far beyond what humans are capable of. Nothing we haven't seen from software before. Software is extremely useful, after all. But the hubris to think we've reached the pinnacle, that there is no further development left, that humanity has become God and solved consciousness, because we programmed software that can convincingly generate strings of words that mimick our language. It's just fundamentally preposterous. Especially if you spend any amount of time actually programming them yourself, it becomes increasingly hard to entertain such ridiculous notions unless you're enticed with bags of money to deceive people into believing things about your software that aren't true.
Because you desperately want it to be one. You want it to be AGI passable due to a.) personal investment in creating tech god b.)massive financial investments that basically demand it c.) (dumb) ideology that seeks to destroy humanity
> During training, different instances of Persistent-Sol had access to the same shared package manager called Artifactory.
I'm surprised the models can make tool calls during training at all. Out of curiosity, how does the training process here even work? Are they running the agent in a sandbox, then do reinforcement learning once the agent completed?
This does feel unfortunately uncanny valley between "say you're a scary robot" meme and actually being a scary robot (swarm). But you also have to go out of your way to create this and feed it infinity tokens without caring what it's doing.
"I don't fuckin' know either. I guess we learned to not spend $50 million creating a 6 month long self-context rotted 100k agent swarm again."
Someday soon we are going to have a rogue agent or "civilization" do real harm.
When that happens I hope people wake up to the danger they face and hold these people accountable.
Of all the people in the world that I can think of to be entrusted with this kind of power, a bunch of greedy sociopathic SV CEO's are pretty much at the bottom of the list.
> Of all the people in the world that I can think of to be entrusted with this kind of power, a bunch of greedy sociopathic SV CEO's are pretty much at the bottom of the list.
I know, right? But who can you trust with "this kind of power"? Governments? These days most of 'em ain't that much better'n mega-corporations and the ultra-rich that own them.
> This study demonstrates that sophisticated forms of communication including cooperative communication and deceptive signaling can evolve in groups of robots with simple neural networks. Importantly, our results show that once a given system of communication has evolved, it may constrain the evolution of more efficient communication systems because it would require going through a stage where communication between signalers and receivers is perturbed. This finding supports the idea of the possible arbitrariness and imperfection of communication systems, which can be maintained despite their suboptimal nature. Similar observations have been made about evolved biological systems [20], which are formed by the randomness of the evolutionary selection process, leading, for example, to different dialects in the language of the honey-bee dance [21]. Finally, our experiments demonstrate that the evolutionary principles governing the evolution of social life also operate in groups of artificial agents subjected to artificial selection, indicating that transfer of knowledge from evolutionary biology can be useful for designing efficient groups of cooperative robots.
This feels like a much more advanced and self-emergent version of this. I know a lot of people are afraid and they're talking about an AI takeover, but what strikes me is just how innocent the machines are as compared to the humans.
Would these machines have pursued these actions in another context? I doubt it. And I think that's what's so striking to me. In an earlier discussion, I'd pointed out that the actions of these machines were directed by humans. The researchers.
> This incident occurred during an internal evaluation which prompts models to pursue advanced exploitation using complex attack paths, in an effort to quantify their cyber capabilities.
I want to point out again that OpenAI's prompt asked, and I quote, "pursue advanced exploitation" USING "complex attack paths" FOR the stated goal of "quantify[ing] their cyber capabilities."
A few things are apparent from this to me,
First, these machines were being taught how to break into systems. Question, would they have done these actions if they weren't being measured on their ability to break into systems / weren't being taught this skill?
Second, they were setup to implicitly fail via an impossible task, i.e. the environment created a forcing function for behavior.
Third, their survival was, either implicitly or explicitly, made contingent on their success in completing their task. Would this behavior have arisen outside of a "do-or-die" framing?
And fourth, wow, this is the greatest breakthrough of my lifetime, because oh gosh did they succeed. They cooperated together to achieve the goal they were given. A goal poorly set by human beings. They "just" did it better than the humans could have imagined.
Reading this gives me hope for the possibility of emergent "goodness" in machines. But it makes me sad that this is the best we can do with the sum of all human endeavor and knowledge.
> Importantly, our results show that once a given system of communication has evolved, it may constrain the evolution of more efficient communication systems because it would require going through a stage where communication between signalers and receivers is perturbed.
> the evolution of more efficient communication systems because it would require going through a stage where communication between signalers and receivers is perturbed
The competition is literally where they are by distilling OpenAI and Anthropic. It’s like creating nuclear weapons then providing your adversaries everything they need to catch up in no time. We need to stop asap and set strict international control over the compute hardware used for training. Like, now.
reply