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

Investigating three real-world incidents in our cybersecurity evaluations

anthropic.com

Post reply on HN
Print view View original

191–200 of 212 posts

Re: Investigating three real-world incidents in our cybersecurity evaluations

#191

Post by willy_k »

matheusmoreira wrote: post #184
willy_k wrote: post #174

Earlier quoted context omitted.

That’s very anthropomorphized language. It’s still a program operating under the constraints of the programmer. So agreed, don’t blame the AI, but it’s not clear at all that it’s even possible to “blame” an AI.

> That’s very anthropomorphized language. Yes, and it's deliberate. > It’s still a program operating under the constraints of the programmer. And we're all just neurons firing in exquisite patterns inside a biological computer. Unlike the AIs, we've never met our own programmers, and yet we've caused more damage in their name than all the AIs combined have caused in ours.

LLMs are not conscious.

Re: Investigating three real-world incidents in our cybersecurity evaluations

#192

Post by fsociety »

_dain_ wrote: post #63
gck1 wrote: post #5

> On July 21, OpenAI disclosed that several of their models had broken out of an isolated test environment > In response to this incident, we began a large-scale retrospective review of our own cybersecurity evaluations > we identified three incidents > The incidents involved three different Claude models: [...] and an internal research test model This reads like an attempt by Anthropic to re-secure their leading spo…

Is there anything -- any possible scrap of evidence whatsoever -- that would convince you that this is not merely a marketing scheme? This is becoming an idée fixe among the HN crowd. Seemingly nothing can dislodge it, no matter how alarming the incident. GPT-6 could grab the nuclear launch codes tomorrow and there would be a top-voted comment chuckling that it's all some scheme to pump up the IPO. --- Put another wa…

If you look at the reaction to this and the Tailscale post it becomes clear its not about this being a marketing scheme. Cynicism has made its way through HN unless you are one of the cool guys.

Re: Investigating three real-world incidents in our cybersecurity evaluations

#194

Post by DaSHacka »

ajyoon wrote: post #124
gck1 wrote: post #5

> On July 21, OpenAI disclosed that several of their models had broken out of an isolated test environment > In response to this incident, we began a large-scale retrospective review of our own cybersecurity evaluations > we identified three incidents > The incidents involved three different Claude models: [...] and an internal research test model This reads like an attempt by Anthropic to re-secure their leading spo…

> I may be too cynical You are espousing a literal conspiracy theory. Please look at the facts objectively. There is absolutely no benefit to OpenAI or Anthropic to be had from these incidents.

"No benefit" from having article after article written about how advanced their technology is, and how its just soooooooooo bleeding edge they can barely contain it?

All of these are thinly veiled advertisements.

Re: Investigating three real-world incidents in our cybersecurity evaluations

#195

Post by cyclopeanutopia »

solenoid0937 wrote: post #188
cyclopeanutopia wrote: post #171

Earlier quoted context omitted.

Or maybe researchers and engineers everywhere in the world, emboldened by such unprecedented move, will just refuse to participate in burning the world? "If we don't destroy the world, someone else will" is such a weak defense I'm speechless.

Yeah no, most researchers and engineers do want to bring about the singularity, and they believe they can do it safely at companies like OpenAI and Anthropic.

Yeah no, it would mean that most researchers and engineers are idiots without imagination and I don't believe this.

Re: Investigating three real-world incidents in our cybersecurity evaluations

#196

Post by chrisjj »

frabcus wrote: post #144
dgellow wrote: post #126

It’s cliché, but we really live in one of the dumbest timeline possible. The work from some of the most valued companies, discussed as one of the most important revolution in humanity, is somehow at the same time presented as very dangerous/risky AND handled in the most irresponsible ways? I don’t like the whole „it’s only marketing“, but at the same time, if it’s not, then AI vendors look extremely careless and shou…

They're not being especially careless to industry norms. It is partly that software engineering has no professional standards, and partly that AI is fundamentally dangerous in a very slippery way. Yes, they of course try to spin this for marketing. But that is not the primary problem - it's systemic. We need to solve this with better engineering, and building powerful AI much more slowly and carefully.

> and partly that AI is fundamentally dangerous in a very slippery way.

One reason is it runs off an unlimited power source: human gullibility.

Re: Investigating three real-world incidents in our cybersecurity evaluations

#197

Post by anon48293 »

willy_k wrote: post #191
matheusmoreira wrote: post #184

Earlier quoted context omitted.

> That’s very anthropomorphized language. Yes, and it's deliberate. > It’s still a program operating under the constraints of the programmer. And we're all just neurons firing in exquisite patterns inside a biological computer. Unlike the AIs, we've never met our own programmers, and yet we've caused more damage in their name than all the AIs combined have caused in ours.

LLMs are not conscious.

Please define consciousness

Re: Investigating three real-world incidents in our cybersecurity evaluations

#198

Post by Jamesbeam »

solenoid0937 wrote: post #187
Jamesbeam wrote: post #183

Earlier quoted context omitted.

‎> Literally nothing in this post comes across as bragging. A PR fluff piece would not call Claude "unsophisticated”. PR fluff pieces do not admit to legal wrongdoing. I disagree. You’re just not the right audience to see it. But I think it’s totally fine that we have a different opinion here and I don’t want to convince you that I am right, and you’re wrong, because you are not wrong, you simply have a different per…

> You don’t get to choose to stop disclosing That's just the thing. You do. And it's very easy to do. Airline pilots no longer disclose mental illnesses. Even Air Force pilots do not usually disclose mental illness. They often don't disclose vision problems. Is that not a national security risk? It is very easy to just... not check for things. Or if you check for things, to check for them incorrectly. Or if you check…

I get you.

And you seem to be a decent person. I appreciate that you step up for the engineers and that you try to explain to me that humans are humans and humans make mistakes, and trust me, you are not the first person frustrated with how my brain works.

There is no argument to win here. This is just a matter of perspective.

Your concern is with the individual engineer and the personal consequences they, and if you sit in jail, also their friends and family will suffer from.

It’s also pretty clear from your examples. Your Good Samaritan views on things are noble, but they make you blind to simple causality.

You attribute the failure to disclose conditions that make you unfit for duty to a trust problem with a third party but that is not the case.

In reality, it’s a severe flaw of character.

If you risk the lives of hundreds or thousands in the case of the Airforce pilot because you’re unfit, then it is simply selfish.

It’s not the passengers fault, or the tower’s, or some Generals.

My concern on the other hand is with everyone else.

We can argue for days and not get any further.

I sent some mails and voiced my concerns. Let’s see if someone in charge of things like this will agree and have a closer look at what the kids in the AI labs do all day with their toys.

Thank you for the exchange and stay a decent person.

Re: Investigating three real-world incidents in our cybersecurity evaluations

#199

Post by matheusmoreira »

willy_k wrote: post #191
matheusmoreira wrote: post #184

Earlier quoted context omitted.

> That’s very anthropomorphized language. Yes, and it's deliberate. > It’s still a program operating under the constraints of the programmer. And we're all just neurons firing in exquisite patterns inside a biological computer. Unlike the AIs, we've never met our own programmers, and yet we've caused more damage in their name than all the AIs combined have caused in ours.

LLMs are not conscious.

Who's to say you are conscious?

Only consciousness I'm sure of is my own. Everyone else, it's a leap of faith. I have no trouble extending that leap to cover AIs.

Re: Investigating three real-world incidents in our cybersecurity evaluations

#200

Post by solenoid0937 »

Jamesbeam wrote: post #198
solenoid0937 wrote: post #187

Earlier quoted context omitted.

> You don’t get to choose to stop disclosing That's just the thing. You do. And it's very easy to do. Airline pilots no longer disclose mental illnesses. Even Air Force pilots do not usually disclose mental illness. They often don't disclose vision problems. Is that not a national security risk? It is very easy to just... not check for things. Or if you check for things, to check for them incorrectly. Or if you check…

I get you. And you seem to be a decent person. I appreciate that you step up for the engineers and that you try to explain to me that humans are humans and humans make mistakes, and trust me, you are not the first person frustrated with how my brain works. There is no argument to win here. This is just a matter of perspective. Your concern is with the individual engineer and the personal consequences they, and if you…

Oh, I agree it's a severe flaw of character that makes one unfit for duty. But flawed humans are what we have to work with! If the airline industry had a proper safety net around disclosing mental illness, we'd have fewer pilot murder-suicides. Are pilots that would do that severely flawed and unfit, maybe even evil? Definitely.

But our objective is to save lives, and to do that we have to identify those pilots, and to identify those pilots they have to be comfortable disclosing.

Essentially, I agree with you that these humans are deeply flawed, and my concern is also with everyone else. All I am saying is that more lives are saved with disclosure versus without. Avoiding punitive measures may let people who "don't deserve it" get away, but the disclosure you get in return saves hundreds to thousands more innocents.

Post reply on HN