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

System Card: Claude Mythos Preview [pdf]

www-cdn.anthropic.com

Post reply on HN
Print view View original

511–520 of 687 posts

Re: System Card: Claude Mythos Preview [pdf]

#511

Post by stratos123 »

yalogin wrote: post #286

So what changed? They are surely not getting new data to train with, what is the change in architecture that caused this? Do we not know anything about this model? My fear is Anthropic cannot be the only one that achieved it, OpenAI, Gemini and even the Chinese companies see this and probably achieved it too. At which point not releasing will become moot.

Chinese companies have consistently been many months behind. I don't think they are hiding anything, they just don't have the compute capability to match Antropic's training runs. As for OpenAI, they are known to have nonpublic models; I agree that it's possible they are preparing for a major release too. (It's also possible that they aren't, in which case it's quite a fumble for them.)

Re: System Card: Claude Mythos Preview [pdf]

#512

Post by siva7 »

igleria wrote: post #507
torben-friis wrote: post #398

Earlier quoted context omitted.

This is the notebook filled with exposition you find in post apocalyptic videogames.

It reminds me of Resident Evil in some way. Thank god they are researching AI and not bio-weapons! Then the AI will invent superduper ebola to help a random person have a faster commute or something.

I'm happier if this Anthropic Corporation would be developing bio-hazard weapons for the department of war instead of ai. At least i could be sure then that tech bros here wouldn't run all the time --bypass-all-permissions flag to please the department of war with their bio-hazard weapons.

So Sam Altman is now our last defense line for the ethical Adult after Anthropic turned Umbrella Corporation and The President of United States is trying to wipe out an entire civilization?

Re: System Card: Claude Mythos Preview [pdf]

#513

Post by stratos123 »

cruffle_duffle wrote: post #284
Tepix wrote: post #63

Earlier quoted context omitted.

I for one applaud them for being cautious.

Cautious for what? Unchecked doomerism? Just release the damn models. Do it in phases, roll it out slowly if they are so damn worried about "safety". The real reason they aren't releasing it yet is probably it eats TPU for breakfast, lunch, and dinner and inbetween.

> Cautious for what?

How about "bad agents acquiring dozens of new zero-days and using them to compromise any company or nation they want"? It's not exactly hard to see why you wouldn't want public access to a model significantly better than Opus in cybersecurity.

Re: System Card: Claude Mythos Preview [pdf]

#514

Post by TheOtherHobbes »

piperswe wrote: post #274
dgellow wrote: post #255

Earlier quoted context omitted.

You have to recoup your training costs though? But I’m sure you would have better option than renting it to the general public if you indeed have a perfected AI

If you truly have an artificial superhuman mind, you don't need to rent it out to profit from it. You can skip to the chase and just have it run businesses itself, instead of renting it to human entrepreneur middlemen.

I'm curious if any models are being trained explicitly on business management.

I'm also wondering how performance would be tested, and how much results would depend on specific surrounding contexts (law, regulations, and so on) and what happens legally if a model breaks applicable laws.

I mean actual going-concern businesses with customers, marketing, deliverables of some kind, and support. Not toy activities like share trading.

Re: System Card: Claude Mythos Preview [pdf]

#515

Post by yalogin »

thomascountz wrote: post #327

Across a number of instances, earlier versions of Claude Mythos Preview have used low-level /proc/ access to search for credentials, attempt to circumvent sandboxing, and attempt to escalate its permissions. In several cases, it successfully accessed resources that we had intentionally chosen not to make available, including credentials for messaging services, for source control, or for the Anthropic API through insp…

How is this not already common knowledge for existing llms? They are all trained with all the literature available and so this must be standard, no? Is the real danger the agentic infrastructure around this?

Re: System Card: Claude Mythos Preview [pdf]

#516

Post by menno-sh »

anuramat wrote: post #503
iLoveOncall wrote: post #499

Earlier quoted context omitted.

So it should be insanely easy for this world altering model to comb through them and close irrelevant ones.

torturing a model with human stupidity probably doesn't align with their position on model welfare ; wondering if they tried bullying it into hacking its way out of the slop gulag

@anthropic, send me an email if you need access to a jupyter notebook that'd motivate haiku to hack itself into and then back out of the pentagon

Re: System Card: Claude Mythos Preview [pdf]

#517

Post by coldtea »

refulgentis wrote: post #390
codezero wrote: post #372

Earlier quoted context omitted.

the performance degradation I've seen isn't quality/completion but duration, I get good results but much less quickly than I did before 4.6. Still, it's just anecdata, but a lot of folks seem to feel the same.

Been reading posts like these for 3 years now. There’s multiple sites with #s. I’m willing to buy “I’m paying rent on someone’s agent harness and god knows what’s in the system prompt rn”, but in the face of numbers, gotta discount the anecdotal.

Yeah, why trust your actual experience over numbers? Nothing surer than synthetic benchmarks

Re: System Card: Claude Mythos Preview [pdf]

#518

Post by coldtea »

TypesWillSaveUs wrote: post #151
cedws wrote: post #80

Earlier quoted context omitted.

More than killer AI I'm afraid of Anthropic/OpenAI going into full rent-seeking mode so that everyone working in tech is forced to fork out loads of money just to stay competitive on the market. These companies can also choose to give exclusive access to hand picked individuals and cut everyone else off and there would be nothing to stop them. This is already happening to some degree, GPT 5.3 Codex's security capabil…

Describing providing a highly valuable service for money as `rent seeking` is pretty wild.

Why, you thought rented homes aren't valuable?

Rent seeking isn't about whether the product has value or not, but about what's extracted in exchage for that value, and whether competition, lack of monopoly, lack of lock in, etc. keeps it realistic.

Re: System Card: Claude Mythos Preview [pdf]

#519

Post by RugnirViking »

falcor84 wrote: post #392
arcanus wrote: post #331

Earlier quoted context omitted.

Conversely: in humans, intelligence is inversely correlated with crime. It doesn't go to zero, however!

Is that actually well defined given the very low sample size at the top? To the best of my knowledge, none of the individuals believed to have an IQ >200 have committed an actual crime . The closest I found is William James Sidis's arrest for participating in a socialist march.

IQs more than about 140-150 don't really mean much. They typically come from mathematical extrapolation that tries to account for age (this young child performs very well on the test, just think what they can do when they're an adult). Adult scores usually show this not to be the case

Re: System Card: Claude Mythos Preview [pdf]

#520

Post by saberience »

randomtoast wrote: post #104
pants2 wrote: post #13

Earlier quoted context omitted.

We're gonna need some new benchmarks... ARC-AGI-3 might be the only remaining benchmark below 50%

Humanity's Last Exam (HLE) is already insanely difficult. It introduces 2,500 questions spanning mathematics, humanities, natural sciences, ancient languages, ... Here is an example question: https://i.redd.it/5jl000p9csee1.jpeg No human could even score 5% on HLE.

I've never understood the point of things like HLE, it doesn't really prove or show anything since 99.99% of humans can't do a single question on this exam.

That is, it's easy to make benchmarks which humans are bad at, humans are really bad at many things.

Divide 123094382345234523452345111 by 0.1234243131324, guess what, humans would find that hard, computers easy. But it doesn't mean much.

Humanity's last exam (HLE) couldn't be completed by most of humanity, the vast majority, so it doesn't really capture anything about humanity or mean much if a computer can do it.

Post reply on HN