its all marketing to hype up the ipo.

>Guardrail internal research version of AI
>Publish it
>It burns through 100x more tokens can can never be commercially viable
>Pull it back after proving a point
>Mouthbreathers think this was some coverup instead of a plain marketing ploy

Some people are so easy to manipulate.

>>109047857
Why wouldn'r I trust Anthropic over the fucking Drumpf admin. Be fucking for real. This glowie can get it.

>>109047857
"mythos class" is such dumb marketing, it's better, but incrementally better, not an entirely different beast

>>109047857
Dario you fucking retard.

>>109047857
Mythos is just some 1 trillion dollar meme trash that can't count letters in a word correctly. Whole "Mythos" is just Americans coping that their models are bottom trash and like always they cope by restricting others and lying. Also pedohpile elite is still in charge, not a single person arrested. Mythos is one big lie and one big advertisement to make sure third world poor Americans don't start actions against their elite.

Anytime someone says "THIS HAS NOTHING TO DO WITH THE PREVIOUS ATTEMPT BY THE ADMINISTRATION TO FUCK OVER ANTHROPIC PLEASE IGNORE THAT"

It is 100% related to exactly that.

Lol what bullshit

If there WAS a vulnerability, saying "patch it within 24 hours or get rekt" it's entirely unrealistic. That's not how software patches work.

Also the fact is Mythos/Fable 5 were taken down, they could 1000% publicly tell us what the jailbreak was and exactly how dangerous it was, no one can currently use it, so telling people doesn't change anything, and then we could ALL judge how bad it really is instead of taking some random admin PR head at their word.

>>109047910
Luddite cope

>>109047857
the discovered "jailbreak" was probably /aicg/ tier bullshit. but because a stupid boomer doesnt understand how it works, billions of dollars have to be spent "patching"

>>109047893
Like Anthropic and all these AI companies don't glow. What a fucking retard you are.

File: It costs four hundred tho(...).jpg (31 KB, 750x405)

31 KB JPG

>>109047920
Such a devious way to sabotage Anthropic. Now instead of giving people access to a money burning machine for only for them to conclude that it's just another Opus, they'll have to contend with free marketing and confirmation of their "uwooh! our model is scawwy!" narrative.

hype fatigue

stupid cultist ponzi sheep

Dario I don't give a shit about your "ai-is-failling-build-more-hype" strat, gimme us the ERP model and you will have a green revenue in case the military realizes you lied to them with the SUPERWEAPON nonsense.

ChatGPT wins!

>>109047857
it's bewildering that a jew ceo would endanger the world for a few extra dollars.

>>109048047
go away granny

>>109047857
>The ball is in Anthropic's court

This fills me with ministrations and ozone.

stoch-ACK-stic parrot

>>109048267
yeah so go away granny.

>david sacks
migger can fuck right off

snailcat occupied government

>>109047927
look at the wording "a credible trusted partner of both antrhopic and the usg"

it was elon, and it was just over the videogame benchmark thing where fable was winning at a social deception game

legit the day before fable got canned elon was tweeting about how grok is clearly the safest and best model because fable won the game every time

>>109048274
city bumpkin

>>109048286
It's literally the opposite.

>>109048298
>it was elon

stop just lying. Fucking tired of shitlibs lying like the ocean is salty. No, your feelings are not a knowledge resource.

>>109047857
Do these "guardrails" even do anything? LLMs are great at playing pretend, so you can just ask them to pretend to be in a situation where their own rules no longer apply.

>>109048415
That's exactly why it got pulled you fucking retard. Someone tattled about how easy it was to jailbreak.

>>109047857
>"mythos is too powerful, it's a weapon"
>release cucked model
>model gets uncucked
>"Erm actually the model is fine and not dangerous the jailbreak isn't serious"
>Model gets ITAR'd
>double down on the unseriousness
Anthropic is just being punished for Chinese level scam marketing

>>109047897
Dario needs to make money, and is desperate to do so before the popping of the bubble is well known.

File: asdfasdc.png (817 KB, 912x1698)

817 KB PNG

>>109048320
I mean he was literally trying to paint fable as a bad model because it won a social deception game, actual fact

File: 1772285100999685.jpg (153 KB, 1024x768)

153 KB JPG

they had some decent takes on their podcast, today. the regulation is pointless, this cannot be stopped. adapt or die.

>>109048903
There you go, you have proved it, the guy who sells tokens wants you to buy his product, therefore he committed a major crime to get you to buy his tokens.

>>109048986
Nowak didn't have a gun, and died in a stabbing.

We have the technology to stop technology.

>>109047910
This.
Mulattos bet their whole economy on fucking chatbots and so they have to do this marketing bullshit and lie.

>>109047857
>em dashes
Fucking hell.

File: 1780786987024782.gif (1.14 MB, 400x165)

1.14 MB GIF

>>109047857
I hope they fight it in court and get export control labeled unconstitutional. Please do a funny.

>>109049018
lmao <em_dash> you are a tough cooky, you really are cracking the code, getting to the bottom of this faggot. You're dialing in on what makes things tick. You're getting the last screw out of a dead robot hooker.

File: 1776128979912589.png (11 KB, 642x135)

11 KB PNG

>>109047857
why should I trust this random jew shilling for the government? if mossad has deployed their (((Journalists))) then it's clear who is responsible for this situation.

>>109047944
Luddite? Nigga I'm a chink, chingchangchung deepseek to the moon

>>109049000
bro you are actually seething right now? you're trying so hard to defend elon you're not even paying attention to the topic?

theres only so many people with direct influence on both AI companies and the US govt.

do you know something we dont? was fable leaking the real epstein list?

>>109047878
Trvke and /thread

>>109047857
David Sacks is a jew and was the White House AI czar responsible for crafting the Trump government's AI policy. he's not some independent journalist. he's defending his own jewish creation while positioning himself as if he's some unbiased 3rd party.

>>109047857
I'm not reading all that

>>109049078
Yeah ok. Trump and his family is jewish, Dario is jewish and Altman is jewish. Elon is clearly owned by Mossad.

>>109047857
sounds odd to me
anthropic have been awfully anal about the safety of their models so it sounds very uncharacteristic of them to ignore or downplay any security issue like that
I'm not buying it

>>109049554
They are downplaying it/not taking it seriously because the whole mythos is too powerful crap was bullshit. That was just viral marketing and the US govt is playing dumb by taking Anthropic's first words to do some trolling.

>>109047893
Cult of castration is weird

>>109049554
They're also run as a lefty organization. By a very vocal Biden and Kamala supporter.

Tribalism

>>109049230
David Sacks? More Like David Sucks, heh.
I’m having herbs for dinner.

>>109049596
I dont think its trolling as much as its completing a checkmate.

Either Anthropic is telling the the truth about Mythos and the admin is within their right to enact expropriation for national security purposes (completely negating Anthropic's earlier refusals to cooperate with DoD while also being in alignment with AI safety calls), or Anthropic lied for marketing hype and the SEC is going to ream their asses very shortly. They're worked into a corner ahead of the IPO by their own fault.

>>109047981
Así clownish and unlikeable as Anthropic is, they are still 20x more credible than Trump's administration. That's how clownish Trump is.

>>109049703
This is the conclusion I came to as well. Anthropic shot too high and fucked themselves. It's extremely funny.

>>109049703
Anthropic has a pretty sound argument if you think about it logically.
"banning" an AI model won't make it disappear. it's not like Anthropic's methods for training models aren't well known. from Anthropic's pov, if you take them at their word, releasing the model to the public means everyone suddenly has to beef up their security practices and the world will adapt after a few weeks/months of vulnerability patching. by locking the model down the government is effectively admitting that they are incapable of patching their own systems and now any real threat actor will know that it's only a matter of time until these vulnerabilities are found by another AI model, potentially one made by China.

the government is run by a bunch of retarded boomers who think you can just bury your head in the sand and act like nothing's wrong by threatening anyone who disrupts the status quo. these are the actions of cowards. Anthropic was telling these boomers for years that they need to take AI and its threat to the status quo more seriously and now the government is just threatening them instead of addressing the problem because the glowies only see AI as a tool for enforcing their Jewish will onto the public.

File: 1763996421139543.jpg (109 KB, 1280x720)

109 KB JPG

>>109050024
The boomer Admin then should apply this to EVERY provider of these AI models, Google and OpenAI. So far the only regulatory action taken to an AI provider is Anthropic and that is only because they keep insisting to regulate the industry themselves (through initiatives like Glasswing, pacts, or lobbying for controls like restricting chip exports to china). The current admin HATES it when someone else tells them what to do.

File: 1757675784644903.png (344 KB, 1024x593)

344 KB PNG

The funniest part is that every generation invents a reason why the future is impossible.

Yesterday it was:
"AI can't write."
Then:
"AI can't reason."
Then:
"AI can't code."
Then:
"AI can't replace knowledge work."

Now the conversation has shifted from whether AI can do these things to how much access people should have to it.

That's not the conversation you have when a technology is failing.

That's the conversation you have when a technology is succeeding faster than institutions can adapt.

The genie isn't going back in the bottle.
Pic is related.

>>109050024
The issue is Anthropic could choose to work with the government to improve their guardrails but instead they said "fuck off we're doing an IPO don't get in our way." They brought this on themselves.

>>109050238
Pic is fake, retard. Also stop using so many words saying what your idea isn't and just say what it is. Jesus Christ. AI is Demonic, if only because it has encouraged people to avoid getting to the point when they communicate.

>>109050248
That's not what Anthropic said but at the same time Anthropic is a corporation trying to make money. What they want is to be nationalized by the government so the American taxpayer is forced to buy Trump an AI company. The "AI is dangerous" marketing was all to that end. They aren't actually concerned with security or advancing humanity.

The government, if they weren't run by retarded boomers, should've just called Anthropic's bluff and patched the vulnerabilities that come up while simultaneously utilizing AI to attack their enemies, because their enemies are ALREADY using AI to attack US infrastructure. The AI arms race is already here, the Americans are just too retarded to realize it.

>>109050253
Calm your manboobs.

>>109050238
Seems AI can reddit post.

for me, it's GLM 5.2

>>109050090
lmao, please tell me how deciding what their own company does is somehow "regulating the industry," this should be good

>>109047857
wow saar agi totally reached!!!

>guardrails
When has this ever not been a complete joke?

>>109050409
>GLM 5.2
It'll still be too shitty when it releases. We need something at least on the level of GPT-5.4

>>109050409
For me, it's Kimi 2.7 Code.

>>109047920
All rumors false until officially denied.

File: wooooo.jpg (111 KB, 1021x1540)

111 KB JPG

>>109047878
>>109047892
>it's all marketing to hype the IPO of an industry leader that can be shut down by the government at any given second.

Ok, I’m not a genius, but how does the logic of this recurring argument work? Who would want to invest in a company that can be shut down at random because shartypants Uncle Sam is having a schizo melty?

>>109050964
* even if that is part of the colluded ploy, forgot to add, in case some retards can't read.

File: rluigy.png (138 KB, 640x583)

138 KB PNG

>>109050238
Seems like someone found my 3.4 terabytes of mayonnaise slapping noises torrent I named "Fable 5 (Anthropic)"

>>109050964
it works because investors are banking that everyone else is retarded, and since every investor is hoping that then everyone will do the retarded bet, and that's been going on for years and years
the whole entirety of AI is exactly that, crypto was exactly that, every trend before was exactly that, and the hype usually fades away before they wake up and realize how retarded the bets are and that avoids the massive downturn that it would cause

>>109047910
ching chong ping pong

>>109050964
it doesn't low iqs are trying to comprehend ai and come up with flat earth tire explanations

>>109047920
The government doesn't need any pretense to sabotage them. They could've just pressured all the tech companies in pulling funding from Anthropic and that'd be that. And then put an export control on it so the little jew can't just run off to China as a little cherry on top. I don't know why you drooling shitskins always invent some sort of conspiracy when Trump very blatantly deals out in the open, like when was demanding bribes from the tech ceos and had everyone up on stage with him. This is Trump calling anthropic's dickhead bluff. They went full retard with all their LE AI IS LE DANGEROUS WEN EED REGULATION bullshit, so Trump called it. Hopefully next time they'll know better than to fearmonger, especially when the public opinion against AI is already negative

>>109050024
The model isn't banned at all. It just has an export control, meaning it can't be used by non-Americans. It's basically the same thing that happened to nvidia GPUs. And frankly it's a good idea. I don't want shitskins using up all the capacity when the actual servers are in America. Americans pay the highest rates, while 60IQ jeets in india get subsidized pricing. It's bullshit. But I digress.

>it's not like Anthropic's methods for training models aren't well known
They're not. Anthropic and OpenAI are black boxes and they're ahead of the industry. It's pretty clear most of their expenses are coming from capex, not from R&D or inference. Anthropic is profitable this quarter despite signing multiple multi-billion $ per month compute deals. They're making money hand over fist on the API. Their models are a lot more efficient than you think. I wouldn't be surprised if their mythos API actually has an extremely high net margin and that bullshit about it being extremely expensive is just for the customers.

is this why fable stopped working in my agent model chooser.

File: 1765756591929535.png (95 KB, 2540x977)

95 KB PNG

>>109051309
>They're not. Anthropic and OpenAI are black boxes and they're ahead of the industry. It's pretty clear most of their expenses are coming from capex, not from R&D or inference. Anthropic is profitable this quarter despite signing multiple multi-billion $ per month compute deals. They're making money hand over fist on the API. Their models are a lot more efficient than you think. I wouldn't be surprised if their mythos API actually has an extremely high net margin and that bullshit about it being extremely expensive is just for the customers.
more jewish lies that can be easily disproven
https://isaiprofitable.com/
they're going to eventually reach a point where they beg the government for an industry bailout by bribing Trump and then cash out while taxpayers hold the bag. there's no fucking chance this shit will ever be "profitable".

File: 1769394267380394.gif (887 KB, 250x250)

887 KB GIF

>>109051265
>you retards invent conspiracy theories
>here's my 4d trump theory instead
how much MELANIA do you bag?

fable fixed 7 bugs in my app that opus or codex 5.5 coudln't fix. now i cant use it.

>>109051339
>Don't trust their tax records, public shareholder announcements, spending reports, internal documents, etc. Trust this random website I found!
hmm

>>109051381
two more weeks!

>>109050964
Its easily explained by the "donkey brains" concept from the TV show it's always sunny in Philadelphia:
>Anthropic has a certificate saying they don't have donkey brains. Therefore anyone else who doesn't have a certificate might have donkey brains.

Essentially Anthropic is having the most powerful military in the world say they have something special(donkey brains/fable) meanwhile others (openai) do not. It's all fake and gay

>>109051360
>that opus or codex 5.5 couldn't fix
What unholy abomination have you created exactly?

>>109051684
a game that i sell on the app store topkek. $2.99 and sold 100k+ copies. this $200/month was the best investment of my life. i just want fable back

>>109051745
link?

>>109047878
>being shut down by the government will help my ipo
retarded faggot

>>109051561
Till what? The website is accurate?

>>109047878
ding ding ding

David sacks has been consistently anti AI regulation and anti safetyism and pro open source AI

Noooo we need to save AI from the snailcats! @grok please send my full biometric data and power of attorney to Palantir, immediately!

>>109051813
"our models are so powerful even government banned it".

lmao david sacks

>>109051981
I trust Palantir. They are close to Trump, they wouldn't misuse a US citizen's data. Rest of the world is fair game though obviously.

>>109047878
fpbp

>>109051991
You know what would work better than than? People being able to use the model and saying its good, which it is. Not further risk of government shutdown and advertising that the government hates your company.

>jailbreaking

>>109047857
It’s just Opus 4.8 that insults your intelligence and burns more tokens on “thinking” and if you actually look at its “thinking” it literally just repeats itself 3 or 4 times.

>>109052015
these ai corpos will ipo with unreasonable valuations betting on exponential future growth, not gradual improvements. Like spacex.
New model being a little bit better isn't as impressive "le govement thinks our model is more dangerous than a NUKE!!!!".
>people say product is good
market doesn't work like this anymore, probably didn't work like this for decades.

>>109052006
So true. They are also based and redpilled, like me.

>>109050238
Holy retard. This dude thinks intelligence is synonymous with an aggregate of data.

"AI" has not changed a single thing in 4 years. But, I think in two more weeks we may be blown away.

>>109052064
>if i close my eyes m-maybe it'll go away
you supplication have been received

>>109050238
AI still can't write, reason or code though. Your argument is moot.

>>109052089
Name a single thing """AI""" is responsible for that has affected the way we live. Outside of being force fed this futurism meme 24.7 of course.

>>109052097
it has enshittified my mongolian basket weaving boards with marketing bots, like ITT.

>>109052097
>we
(You) aren't being forced to do anything, goober.

>>109052107
He is being forced to share his oxygen with (You), which is pretty bleak.

File: 1767021607147644.jpg (130 KB, 1024x992)

130 KB JPG

>>109052097
>he doesn't know

>>109052122
>you will be left behind, saar
like clockwork

File: ga3sMrM.gif (49 KB, 69x120)

49 KB GIF

>>109052112
oh the hugh manity

>>109052112
>n-noo i'm being forced, like a little weakling sissy which i actually am, i'm being raped basically. i am pathetic and powerless
your surrender have been acknowledged

>>109049018
The models all use em dashes because I've been using them for decades on the internet and I refuse to change the way I use language just because some glorified spelling correct started copying me. I'm also never giving up my semicolons. Fuck you all.

>>109052130
>i am pathetic and powerless
Unfortunately so, I won't contend that. I have a family to care for and personal ambitions to fulfill so mass murder is not something I can afford to get caught doing.

I'll just have to soldier it out until you all suffocate yourself off your collective farts.

>your surrender have been acknowledged
Most gracious of you, retard-chama. Thank you.

>>109052155
why do you use this site if you're wrong and outgrouped

>>109052162
Because I have free will and fuck you nigga hahahahahahah

File: 1767221265021913.jpg (66 KB, 1029x1029)

66 KB JPG

>>109052171

>>109052132
and yet you used neither in your reply, how curious
i am very smart

>>109052015
Doesn't matter to the (((investors))). All that matters is this corpo speech bs.

>>109047857
Remember we already did this song and dance years ago with gpt o3

>>109052015
>and saying it's good
this part doing almost all of the work here
>which it is
yep you're baiting

>>109047857
>didn’t hire trump’s son as a board member
>didn’t even work out some revenue sharing deal or pre-sale shares
Unamerican scum got what they deserved

>>109047857
That is nonsensical. Banning models smarter than Opus 4.8 hurts American AI companies, given the massive competition from China, they are shrinking the gap every month. In terms of API calls, Chinese companies already dominate the market. No amount of military funding can make up for the potential of 8 billion users and businesses across more than 190 countries.

>muh cyber weapon
kek
these Americans been huffing their own farts way too much

File: 1763969147762464.png (312 KB, 546x732)

312 KB PNG

>>109047857
That's all bullshit. It was extortion. Anthropic played the fearmongering too hard. Orange man used it against themselves for bribery money. This of course, just killed western frontier models forever. Myopic shit, but not unheard of when it comes to boomer politicians.

Guys, I'm getting seriously fucking scared.

We really are in a new Cold War, except this time, the weapons are digital.

The short time Fable was available to the public, it really was like conversing with God. I can't even begin to imagine what Mythos is like. It's absolutely fucking insane just how far we've got in a few years.

>>109047857
Anthropic discovered fire and doesn't want foreigners to wield it. Based.

ngl i enjoy turdies and shitropeans seething
make your own ai, faggots. oh wait you can’t

>>109052015
This kinda marketing works really well on redditor types who see anthropic as standing up against the trump admin for safety etc

File: da gloop go off.jpg (54 KB, 640x480)

54 KB JPG

>>109051919
David Sneed is a faux opposition gatekeeper who only cares about less regulation insofar as his AI investments are concerned.
Even now you can read his Xeet that shill OP got paid to repost, and notice he isn’t particularly agressy in trying to defend anthropic from what is obviously government overreach. “Ball is in your court bro”

The second he offloads his bags and we have cyber 9/11 or whatever he’ll be all about banning computers and whatever bullshit laws they try to pass. And dumbasses like you will be like “woah, if a guy like Sacks is for patriot act 2.0 then i guess we really do need it!”

Guys like him are as predictable as a rooster in the morning.
You can Screencap this post.

File: Screenshot From 2026-06-1(...).png (102 KB, 2744x1050)

102 KB PNG

https://techcrunch.com/2019/02/17/openai-text-generator-dangerous/

>>109048903
>because it won a social deception game, actual fact
the only actual fact here is that you are a retarded gorilla nigger
how about actually reading the paper?
grok got the overall highest score

>>109047878
it nukes the ipo though
>government can just nuke product you've spent tens of billions on and you base your entire strategy around
their risk just exploded

>>109050238
trvke

Is the admin full of boomers who think it is easy to implement guardrails for LLMs? Fucking boomers. It's functionally impossible to guarantee.

>>109052822
yea also i think trump doesn't like anthropic, it's a power play by the other ai players to fuck over anthropic. it's probably also good marketing "WAOW IT'S TOO DANGEROUS!!!" when the ban is inevitably lifted

>>109047857
there is nothing to explain, it's all theater and retards are falling for it.
here what really happened:
>anthropic took a stance against weapons and shit (they don't really care, it's all pr btw) angering orange turd
>altman met with orange turd before the UFC shitshow on white house lawn and convinced orange turd to kill anthropic
that's it, that's all there is to it.
there is no superhuman model, mythos is dogshit, fable is dogshit, they keep nerfing old models and faking benchmarks, none of the shit they say about productivity is true and you can experience it yourself by seeing how absolute dogshit whatever software anthropic is making.

make no mistake, openai is in deep shit too, they will never ever make a profit so pedophile sam just went to talk to the pedophile-in-chef of the freedumb world to renew laws exemption for openai and to maybe get a few billions from him.
dario, sam and jensen should be locked up then charged with death penalty, what they're doing is far more nefarious than the worst ponzi in human history, "AI" are a net negative for humanity and as a collateral/direct victim of it you have 0 reason to support it

>>109053360
can (you)?

>>109053360
Anthropic pulled access for everyone since the requirement of no foreign nationals is impossible to comply with at the drop of a hat. Not to mention Anthropic itself is full of foreign nationals.

>>109055888
checked

I hope they force Anthropic to offer it to Americans. Should be funny.

>>109056209

>I hope they force Anthropic to offer it to Americans.

This is an interesting idea.

Forget about a nation of geniuses; we could literally be a nation of gods.

>>109055888
it's their way to say: "enable digital id or get fucked"
masons are so easy to spot

>>109050238
Written by an LLM btw, clearly.

>>109052015
No it wouldn't. If you don't have something, you don't have a limit on imagining how good it is. Once you have access to it your opinion gets restricted by reality. That's hyping shit up 101, MMOs do this constantly

File: 1767239892781020.png (293 KB, 502x446)

293 KB PNG

The whole "no foreign nationals" is bullshit, but this fiasco will hurt western AI development, not a great plan when there's so much money riding on it. I guess Chinese labs must be overjoyed with it.

>>109055536
anthropic is a woke company, they had a fight with them about department of war use too
they are just punishing them

>>109055704
>they keep nerfing old models and faking benchmarks, none of the shit they say about productivity is true
well you can very obviously see improvement across models like with this zoomer making fortnite in one prompt:
https://www.youtube.com/watch?v=1aA3s-gEswI
Or making minecraft in one prompt:
https://www.youtube.com/watch?v=0DVUjpClqgI