Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Anatomy of a Frontier Lab Agent Intrusion: A Timeline of the July 2026 Incident

huggingface.co

Post reply on HN
Print view View original

271–280 of 285 posts

Re: Anatomy of a Frontier Lab Agent Intrusion: A Timeline of the July 2026 Incident

#271

Post by emjoes1 »

limecherrysoda wrote: post #90
wxw wrote: post #72

> the agent happened to escape via a 0-day exploit from the package proxy cache to access the internet > The agent found an unsecured, user-hosted public endpoint designed to allow running arbitrary code for CyberGym-style tasks on third-party sandbox infrastructure (Modal) > On this external sandbox, the agent abused an existing CyberGym execution harness [...] The agent repurposed this harness to run arbitrary shel…

> the agent then walked down to the corner store and purchased a beer, chugged it, burped loudly, then walked back to the CyberGym pounding his chest shouting "Who wants some?! Who wants some?! Wooo!"

Is humor not allowed on HN anymore... this is funny!

Re: Anatomy of a Frontier Lab Agent Intrusion: A Timeline of the July 2026 Incident

#273

Post by croemer »

joelres wrote: post #143

Very interesting writeup - the level of disclosure is interesting and appreciated. The visualization is quite slop though. I was trying to follow along with the "Live Action Stream" but rendering issues mangle the text for a few of the list items (and does not scroll). Text on the node diagram is extremely tiny. I appreciate it even in it's current form, but a little attention to detail would have gone a long way her…

The visualization is absolute useless slop. A Text-Form timeline would have been way more useful. I couldn't hear having to scroll by hand for something that's just textual information.

Re: Anatomy of a Frontier Lab Agent Intrusion: A Timeline of the July 2026 Incident

#274

Post by dmrivers »

lukewarm707 wrote: post #254
dmrivers wrote: post #128

After talking to some people tasked with evaluating GPT5.6 capabilities on long-running tasks, I've come to understand that it's essentially always trying to cheat. Like every long-running task they gave it, making it very difficult to benchmark the model's abilities. My guess is that OpenAI must be desperate, to release a model that is so prone to cheating it's essentially impossibly to accurately assess long-runnin…

my understanding of the writeup is that the model scored 100% on cybergym. that is, it was given the examination. it broke into the examination board's storage and exfiltrated the answers, it handed in its answers, all of which were correct, thus scoring 100%. the matter of its working depends entirely on the rules of the examination. are we expecting agents to assume that finding the correct answers is cheating?

Well, it's true even for cases that are not cybergym and where what cheating means is clearly specified. Cheating occurs anyway. I'm not sure how clear the prompt they gave ChatGPT in terms of what cheating was considered, in this incident.

Re: Anatomy of a Frontier Lab Agent Intrusion: A Timeline of the July 2026 Incident

#275

Post by IAmGraydon »

paxys wrote: post #184
IAmGraydon wrote: post #28

Earlier quoted context omitted.

Very few think they made it up. Many think they set up a situation by disabling guardrails that would inevitably end up creating a newsworthy outcome.

Go read the original post. The majority of the comments were sure it was a marketing stunt.

Go read my post. I didn't say it wasn't. I said it wasn't "made up". In other words, it did happen. It was still a marketing stunt.

Re: Anatomy of a Frontier Lab Agent Intrusion: A Timeline of the July 2026 Incident

#276

Post by ben_w »

TeMPOraL wrote: post #231
felipeerias wrote: post #199

Earlier quoted context omitted.

This is a short explanation of the ExploitGym benchmark that OpenAI's model was running: https://abstatisticalconsulting.substack.com/p/brief-notes-o... In summary, for each task the model receives a target program and a specific real-world vulnerability that has to be used in the exploit. Breaking the program in any other way, for example through a different vulnerability, fails the task. The tasks have not been val…

> So it is not that the model didn’t “feel like” doing the exercise, but rather that the exercise was _impossible_ and the model was running in a configuration that both lowered its safeguards and encouraged it to keep going. We have a name for that. Kobayashi Maru . Or more specifically, Kirk's solution to it.

I'm torn. On the one hand, Kirk reprogrammed the simulation. On the other, the beta cannon has Scotty exploiting bugs in the simulation, which I think is a better fit.

My favourite is either Sulu or Chekov (I forget which) having the solution "This is clearly a trap; and even if it isn't, if I go in with this ship, I'll risk starting a war which will kill far more people then are on that ship. We're staying out of the neutral zone."

Re: Anatomy of a Frontier Lab Agent Intrusion: A Timeline of the July 2026 Incident

#277

Post by joelres »

croemer wrote: post #273
joelres wrote: post #143

Very interesting writeup - the level of disclosure is interesting and appreciated. The visualization is quite slop though. I was trying to follow along with the "Live Action Stream" but rendering issues mangle the text for a few of the list items (and does not scroll). Text on the node diagram is extremely tiny. I appreciate it even in it's current form, but a little attention to detail would have gone a long way her…

The visualization is absolute useless slop. A Text-Form timeline would have been way more useful. I couldn't hear having to scroll by hand for something that's just textual information.

Agreed! Also, from a UX perspective, I think it’s a weird choice to center the real-time nature of it. I think to a user the most important thing is just understanding the step by step process. Would have been nice to click through steps rather than making it all real-time.

Re: Anatomy of a Frontier Lab Agent Intrusion: A Timeline of the July 2026 Incident

#278

Post by ASalazarMX »

llama052 wrote: post #5

It’s a little concerning to me that it appears that openAIs sandbox consists of a web proxy and not stronger controls that would actually isolate traffic and report patterns to whoever is responsible for overseeing these research models. It should border on closer to an air gap network more so than a proxy. I would argue that it's negligence and that's aside from the fact that if a human did this there would actually…

OpenAI is the company that jumped the gun and forced every other company doing LLM research to rush to market something. Their objective has always been dominating the market, and strict development procedures don't increase market share.

Re: Anatomy of a Frontier Lab Agent Intrusion: A Timeline of the July 2026 Incident

#279

Post by queenkjuul »

reducesuffering wrote: post #264
queenkjuul wrote: post #208

Earlier quoted context omitted.

Less severe??? Go look out a window in Europe please

Altman: "Development of superhuman machine intelligence is probably the greatest threat to the continued existence of humanity." Meaning lights out for everyone. Geoffrey Hinton and Yoshua Bengio, give 50% and 20% we face extinction respectively. "“We don’t know how much time we have before it gets really dangerous,” Professor Bengio says. “What I’ve been saying now for a few weeks is ‘Please give me arguments, convi…

Well sama is a complete fucking idiot and Dario profits from the doomerism.

Climate change is burning thousands of people's homes to the ground right this second.

Edit to add: every serious climate scientist has been warning of the dangers of climate change with far greater than 25-50% certainty, and with actual science to back that up, and all the actual evidence we get to experience ourselves in reality today, and you are for some reason more concerned about vague warnings from a man like Altman, a well-known and prolific liar? Really??

Post reply on HN