Comment by maebert
7 hours ago
The whole artificial scarcity Anthropic created around Mythos / Glasswing is quite brilliant to be honest (I’m Not saying ethical, just brilliant). The commercial gains are one side of course. But consider this:
Gets labelled supply chain risk by the pentagon. Hypes up what they claim to be the most advanced hacking tool on the planet. This puts the US government into a loose / loose position. Either deny the NSA access to it, or be called out on their bluff.
> The whole artificial scarcity Anthropic created around Mythos / Glasswing is quite brilliant to be honest
Isn’t that just the same strategy OpenAI has used over and over? Sam Altman is always “OMG, the new version of ChatGPT is so scary and dangerous”, but then releases it anyway (tells you a lot about his values—or lack thereof) and it’s more of the same. Pretty sure Aesop had a fable about that. “The CEO who cried ‘what we’ve made is too dangerous’”, or something.
https://en.wikipedia.org/wiki/The_Boy_Who_Cried_Wolf
Right, but in Aesop’s fable, the wolf did eventually come. It’s asymmetric, because in this case the wolf is not coming for the boy, it’s coming for everybody else
The boy isn't crying wolf strictly to save himself. He does it to get the attention of the town, knowing they'll come to the aid of the livestock he's been tasked with watching. Yes, their aid is primarily to save the boy, but the danger is still to the larger community rather than isolated to the lookout.
They way they've published hashes of the bugs it has found so that once those bugs are fixed they can responsibly disclose them while also proving that they weren't lying... that displays a willingness to dabble in evidence which is far beyond anything OpenAI has done to support their claims.
This. I see much cheap naysaying without referenece to the vuln hashes. If it is smoke and mirrors, then the naysayers should loudly shout down the specific hashes and when they get revealed, or don't, then they will have done a great service to dissuading fake claims to world changing tech.
>Sam Altman is always “OMG, the new version of ChatGPT is so scary and dangerous”, but then releases it anyway
One of the many reasons nobody should give Scam Altman their money. It's continually infuriating that this serial grifter is in such a position of power.
It was from GPT-2 and Dario was part of the developers of that model while he was working in OpenAI, not Sam Altman, it's his playbook
> It was from GPT-2
Prior to the released of GPT-5, Sam said he was scared of it and compared it to the Manhattan Project.
1 reply →
This is pretty much correct, but Mustafa Suleyman has probably been doing it longer.
Not just part of the developers, but rather "led the development of large language models like GPT-2 and GPT-3" as per his website.
https://darioamodei.com/
[flagged]
Anthropic has not in fact released it, and it does in fact appear to be that dangerous, judging by the flood of vulnerability reports seen by e.g. Daniel Stenberg.
Certainly it’s a strategy OpenAI has used before, and when they did so it was a lie. Altman’s dishonesty does not mean it can never be true, however.
The flood of reports that open source projects like curl, Linux and Chromium are getting are presumably due to public models like Open 4.6 that released earlier this year, and not models with limited availability.
How many months till they release a better model than mythos to general audience?
Gpt 2 wasn't released fully because OpenAI deemed it too dangerous, rings a bell? https://openai.com/index/better-language-models/#sample1
1 reply →
> judging by the flood of vulnerability reports seen by e.g. Daniel Stenberg
Maybe I've missed anything, but what Stenberg been complaining about so far been the wave of sloppy reports, seemingly reported by/mainly by AIs. Has that ratio somehow changed recently to mainly be good reports with real vulnerabilities?
3 replies →
Those vulnerabilities were found by open models as well.
3 replies →
> This puts the US government into a loose / loose position.
You might even call it... a tight spot
Side note, how did the word "lose" become "loose"? I've seen this so many times on HN.
It didn't, but the advent of spellcheck and autocorrect has made everyone completely give up on proper grammar or word selection as long as no squiggly line appears.
6 replies →
Because your pronounce them backwards.
"Loose" is a short word that ends sharply, but "lose" is a long word that slowly peters out.
They should be the other way around imo.
15 replies →
I’m guessing most cases of loose/lose switch happen when English isn’t someone’s first language.
4 replies →
And let’s not get started on it’s vs. its-—a distinction that now seems irretrievably nerfed
I always assume not everyone is an English speaker and let it go.
3 replies →
Exactly the same way that the `cancelled` of my youth became `canceled`. By being misspelled so often that the misspelling won.
In this case, it's not clear who wins yet — "lose" may loose, or mount a comeback, resulting in "loose" being the one to lose.
1 reply →
It doesn't make sense to have "lose" pronounced as it is. We have rose, pose, dose, nose all pronounced with ō. And then you have lose pronounced as loo͞z. It feels natural to put two O's in there when you write it.
3 replies →
I try to let it go, but this is my pet peeve.
people are from many places
2 replies →
It's fine, nothing to see. Just focus on the intended meaning not the underlying delivery. Mere words don't really impact communication. Right?
1 reply →
Ok. This is was either brilliant or I did not wake up yet.
This is not the first time Pete Hegseth charged into a bar, started swinging his fists and screaming "don't you know who my father is", only to find his junk in a vise with no graceful way get it out.
For some reason I thought you were doing a setup for a joke...
"The President of the US, the Secretary of Defense, Iranian Prime Minister walk into a bar..."
1 reply →
Mythos is most certainly not hype. I think it might be the agent with most agency as of today (ability to get really difficult shit done on its own). I believe that it most certainly is not hype. A realization just struck me that guarding the model weights (which are probably in the realm of a few TB) should be of utmost importance. Essentially - having access to them and a small NVIDIA cluster is all it takes for anybody to start using Mythos for themselves.
Barring any limitations of my understanding, the Mythos model weights are probably in the realm of a few TB. Any actor with access to the weights + a single beefy NVIDIA cluster and a few intelligent folks is all it takes to gain access to Mythos.
Cost of infra < $5 million (guesstimate). Imagine someone pulling that off by gaining access to the weights - which would be a monumental challenge, but likely less complicated than re-acquiring enriched substances from the gulf nation under attack right now. It would be the heist of the century.
> not hype
Proceeds to write the hypiest comment possible. No substantial claims of why the model is not hype, just how dangerous it would be if the weights leaked and how cheap it would be for anyone to just start using it for EVIL if it ever did.
>pulling that off by gaining access to the weights
This was a point in the AI 2027 videos you see on youtube. That model weights would be a subject of active attack by nation states and that governments would start requiring AI companies to treat them like munitions when securing them.
I'm a crypto wars veteran, discovering the internet with the nerfed 40-bit version of Netscape
> Glasswing
Fun fact, the model isn't quite the important part for Glasswing, someone took the ideas, and made their own open alternative, you can swap out models and find issues in code using clearwing. I haven't had a chance to personally test it, but it makes a lot of sense to me.
https://github.com/Lazarus-AI/clearwing
It is pretty obvious from the token speed that opus now is sonnet or haiku size a few versions ago. So Mythos is likely what was called opus. They dont tell us the size but they did co firm the training run for Mythos was under the 10^26 flops reporting requirement.
In an alternate universe, opus 4.7 is sonnet 5, and Mythos is released as Opus. Can you imagine how much praise would be heaped on Anthropic if it opus 4.7 was < half the price it is now?
They created the model specifically to play this game.
“Show me the incentives and I will show you the outcomes.” Charlie Munger
They said they designed it to be a better coding model. Something that has long been true: better software engineers are better vulnerability hunters as well. I think we are seeing that play out with Mythos.
'Anthropic is / isn't lying about Mytho's capabilities' is the less interesting conversation.
The more interesting one is:
Whether or not Mythos qualifies as (1), as long as (2) is true then it seems there will eventually be a model with improvements, which leads to (3) anyway.
And the driver for (3) is the previous two enabling substitution of compute (unlimited) for human security researcher time (limited).
Which begs questions about whether closed source will provide any protection (it doesn't appear so, given how able AI tools already are at disassembly?), whether model rollouts now need to have a responsible disclosure time built in before public release, and how geopolitics plays into this (is Mythos access being offered to the Chinese government?).
It'll be curious what happens when OpenAI ships their equivalent coding model upgrade... especially if they YOLO the release without any responsible disclosure periods.
> Which begs questions about whether closed source will provide any protection (it doesn't appear so, given how able AI tools already are at disassembly?)
Disassembly implies that you're still distributing binaries, which isn't the case for web-based services. Of course, these models can still likely find vulnerabilities in closed-source websites, but probably not to the same degree, especially if you're trying to minimize your dependency footprint.
You're still at the point that any known or unknown disclosure of your binary puts you at risk. At best it's a false sense of security.
> it doesn't appear so, given how able AI tools already are at disassembly?
If that's your concern, shareware industry developed tools to obfuscate assembly even from the most brilliant hackers.
That's not true, they did do obfuscation but the main sneaky thing they did was to make hackers think that they had found all of the checks, and then hide checks that would only trigger half way through the game. That kind of obfuscation is also not relevant to security vulnerabilities.
AI is already superhuman at reading and understanding assembly and decompilation output, especially for obfuscated binaries. I have tried giving the same binary with and without heavy control flow obfuscation to the same model, and it was able to understand the obfuscated one just fine.
Plot twist it gets acquired by the US govt.
If this happens it's not going to take the form of them getting "acquired", they're going to end up forced to become a defense contractor like Lockheed Martin or Raytheon where their primary customer is the USG and all of their sales require governmental approval.
And the absolute last group the government would ever approve access to would be "We the People".
I know it's not realistic at this point, but I really hope the Chinese labs will release models that run local and are on par with the abilities of frontier models. That is, I hope the idea of frontier models goes away. Because if not, what we're looking at is a seriously bleak outlook with respect to economic freedom for anyone outside the 0.1%. We may even be looking at out and out lack of economic viability for vast segments of the population.
It's like opening up an exclusive night club. Everyone is talking about it and wants in, even though most know nothing about what's actually inside.
I'm kind of surprised that C-suite folks fall for this marketing ploy when many of them are typically very close to the sales process in very high stakes areas. I guess it just shows you that anyone is susceptible to a well done grift. On second thought I'm thinking back through the history of C-suite decisions I've seen first and second hand and I'm not surprised at all.
I'm really tired of these claims that Mythos is "nothing by PR hype". It should be at this point eminently clear that the people working at Anthropic believe the things they say about their models. And for mythos in particular, at this point there are far too many people outside of Anthropic who have seen it and/or the vulnerabilities it has discovered for "it's nothing but hype" be anything close to a sensible position. I'm not saying we should blindly believe them; they have often used more caution than was entirely warranted (this is, in my opinion, a good thing) but the idea that all of this around Mythos and glasswing is nothing but marketing hype is nonsense. Might a disinterested 3rd party decide that they think the fire is smaller than Anthropic's smoke warranted? Yes that's possible. But the idea that it's all smoke and no fire at this point deserves no resepect whatsoever.
To be clear I’m not claiming that Mythos is _nothing_ but PR hype, merely that Anthropic is playing its cards really well, which is a claim independent of actual capabilities of their latest model.
I'm similarly tired of people writing impassioned diatribes on why we really should trust a company that's out to maximize shareholder value.
"It's so dangerous that we'll only release it mostly to the companies that have some financial stake in our company"
We don't owe anthropic anything, including benefit of the doubt. They're here to sell products, any other mission statement is a convenience for them.
Worth noting that Trump was one who labeled them a supply chain risk for the horrible crime of setting really basic guardrails around usage. (And it's "lose" btw)
Governments are sovereign: they tell people what to do (by making laws, by exercising a monopoly of violence, etc), and nobody tells them what to do. Governments also fight wars, which means lives depend on the government's ability to command.
Private companies make products. When those products were plowshares or swords or missiles, the company didn't really have a say over how they were used, and could be compelled by the government to supply them. Now that new cloud and AI products that increase government command abilities live on servers controlled by private companies, private companies think they can tell government what to do and not do. No government will accept that, because the essence of government is autocratic sovereignty: the sovereign commands and is not commanded.
In American law, companies have the choice of whether or not to do business with the government, outside of a few corner cases. There’s a process for forcing them, but it can’t just be because the leader says so.
In this particular case Anthropic had a contract stating what the military could and could not use their models for. The military broke that contract. Anthropic declined to sign a revised one.
This is within their rights, and more to the point, the government should absolutely not be allowed to unilaterally alter contracts they’ve already signed!
Predictability is the whole point. Undermining it is how you destroy your own economy.
7 replies →
> the essence of government is autocratic sovereignty
*was
Democracy was and is radical for putting the common people in charge of the government. The right to petition for redress of grievances is literally in the first amendment. Government is a social contract, enforced with state violence on one end and mob violence on the other.
If you want to return to autocratic rule, I hear North Korea is lovely this time of year.
3 replies →
turns out it was spelled "lusage" the whole time
"basic guardrails" within activation capping is not separable for high granularity trained models. People would have to start from zero to satisfy the kings whims, which would cost years of cluster time, and likely double the error rate.
Governments are difficult customers for software firms, as most military folks get an obscure exemption from copyright law at work. Anthropic finding other revenue sources is a good choice, if and only if the product has actual utility (search is an area LLM are good at.) =3
The position doesn't matter. Nobody sane listens to what the orange or "the USA" says because it could be the complete opposite tomorrow. Which sadly is exactly the position where the orange wants to be. Free reign for him and nobody cares.
I think the Dutch would take issue with you throwing around "orange" like that.
If Alexander or any of his usurping ancestors has a problem then he can go ride a horse over a molehill. Oh, what, is that line a bit too soon? Tandem Triumphans!
Not only that, but I feel there's a lot to validity of this meme from reddit: https://i.redd.it/jxfayl16q5wg1.jpeg .
Maybe not "completely out", but at least not having enough available capacity to release a model way bigger than Opus publicly.
[dead]
[dead]
> The whole artificial scarcity Anthropic created around Mythos / Glasswing is quite brilliant to be honest (I’m Not saying ethical, just brilliant). The commercial gains are one side of course.
You mean the obvious commercial losses caused by keeping an expensively created product effectively off the market altogether?
What the actual fuck is with people who come up with stuff like this?
I think Dario didn't get a Gmail invitation back in the day, and now he's taking it out on everyone.
I'd be okay with our military / NSA having the best model possible.
Now if only the NSA would vet key people in our government, there should be no reason a foreign entity can just hack the FBI director's personal GMAIL, the NSA should be trying to break into their accounts before our enemies do. It's ridiculous that they're not already doing this.
>Now if only the NSA would vet key people in our government
They probably did that for a while.
Sadly, they as an agency were un-vettable to the general public, and abused that position to create tons of blatantly unconstitutional programs that they tried to hide.
I agree, I know some people hate the surveillance stuff, but unfortunately we only hear the bad mostly of what it does, we never hear the actual good impact some of these agencies do. I wish they'd release some sort of annual report, but how do you do that without telling your enemies that people are "trying" or being "caught" doing things. It's a pain in the butt.
There are truly evil people in this world, way worse than we probably realize. Our military is not perfect, our country is not perfect, no country or military is, but we generally do our very best to do what is right historically speaking. It's hard to see that if you get lost in the politics of things.