Donald Trump says you're not allowed to use Claude Fable or GPT 5.6
They spent the last 18 months playing up the national security implications of AI chatbots and now the US government took the bait.
Based. You shouldn't be using any gAy I.
>>109135403snailcat moment
>>109135387Local hardware just got more expensive
>>10913538710 years of unregulated AI
very dumb
>>109135387Total snailcat victoryHail Trump
>>109135387well done you fear mongering FAT BITCH
>>109135387china won
>>109135387Is this what "economic freedom" looks like?
>>109135598No, it's what very high-tier stock manipulation and collusion looks like
>government will approve each user individuallyDMV queues on steroids niggas
>>109135387I wonder how you get approved by this just and faithful administration.
>>109135718If we convert to Judaism you can skip to the front of the queue btw
>>109135718>implying it would be any different under a different administrationlol and also lmao
>>109135731>convert to JudaismGoyim... I
>>109135387and that's a good thing
>>109135740Most jews are converts, even Trumps daughter converted and married into a powerful chabad family.
So if these dipshits can't even paint a pool properly how are they going to evaluate frontier models with any efficacy?
>>109135387>the government will approve each user individually
>>109135387lol
>>109135761>mfw gpt 5.8 has released and i'm still somehow waiting in line for 5.6 Sounds fucking awesome
> Donald Trump says> Stepani Palzzozzo> Andrug Curry> Abortimus> and other retards
>>109135732Can't remember any other Presidents that started their term off with a crypto rugpull.
If I had told you ten years ago that text prediction algorithms trained on publicly available data would bring about the end of humanity, would you have believed me
NYT: OpenAI Leans Toward Waiting Until Next Year for I.P.O.
>>109135387Trump actually fell for the bait kek
>>109135822That nigga is old, he's solely focused on enriching his family at this point.
you stupid ass millennials memed yourselves how do i open pdf boomers are now death scared and they have all the power to stop you which they are lmao
>>109135704this is corpos you're talking about they run the government anon if they apply before 11 AM they will be approved by lunch
>>109135387
>>109135387All part of the marketing
>>109135387Welp, China is gonna party hard.Move slow, snailcat!
>>109135898>>109135908They are gonna shut it down once their models reach fable capacity. Screencap this
>>109135456>Heil Israel
>>109135387I said not even 3 weeks ago that the US would regulate and ruin AI because the US has a long and storied tradition of blowing its face off with a shotgun every time it builds an unassailable industry, and /g/ told me it would never happen
>>109135819lol this is FOR the ipo. models so good and scary the government wont let us release them yet. What do you think that skepticism will do to the price? After IPO openai and anthropic will PUMP HARD since that mystical too good model will be in their pockets unreleased
>>109136729Limiting the userbase is gonna drain capital, are you fucking retarded or something?
>>109136780They were never profiting to begin with?
>>109136780The very fact that you oppose this so much makes me think I'm onto something.
This may unironically pop the bubble and lead to a world of self hosted open weight Chinese LLMs, which would all be based.
>>109135412>>109135456>>109135908total putin victory
>>109136710Working as (((intended)))Same old story
Anthropic work for China.
>>109135387Handing free wins to China.Would be a great time for Mistral to release a Mythos class model right about now, too.
>>109135387going all in on chinese models thenget fucked gweilo
>>109135387This is the perfect time for local AI to exceed these fucking "frontier" models. This will also help the chinese release new better models. Shit is fucking retarded and pointless.>>109135403Im sure you use it zoomer faggot
>>109135731So be a christian? Thats basically what you are saying you retarded fuckhole.
The short term greed is impressive. It can't work though since Chinese models are totally viable alternative. I wonder if they'll start war with China over the next year.
>>109136710Trump has been nothing but a corporate whore his entire political career, and now hes sabotaging the largest (and only) source of economic growth since covid? This only makes sense if Israel ordered him to do it.
>>109135397Their new models are shit, so they need some sorry ass excuse to tell their customers they can't use them.
>>109135387As an American, I have an unailienable right to Fable.
>>109135761Yes. It's called capitalism, according to republicans. If you are against the government approving your use of tools, then you might be a communist. If you are against price setting according to the government, then you might be a communist.
>>109135387Because they can basically oneshot the programs the billionaires want to push as a service, security my ass.They've all realized the monster they sacrificed everything to push to fruition is capable of just cloning a lot of their services in one prompt so they don't want access to be so widespread. Thats the real concern, GLM can probably tell you how to hack into most generic vibe coded website databases anyway.
>>109135387Americans are, these retards just won't ID check the chinks for some reason
>>109137326It's not that simple. Only a real cretin would ever say anything like this. Please take a look at **e**e**e's Odysseys "project" for more information. You ain't one shotting shit unless it's a demo what will impress retards.
>>109135387Real life is very quickly becoming science fiction. As a software engineer it's both terrifying (for my employability prospects) and obscenely interesting at the same time.
>>109135662Let me guess, they are going to get away with it?
>>109136687Like they shut it down when they reached GPT 4 levelI can see your kippah
>>109135506this
>>109137402You didn't stand up for racism. Enjoy fascinating being laid off forever.
>>109135387>big US corporations get approved immediately>smaller businesses take a few months>individuals? maybe when the next model is out :)Regulatory capture
>>109135506>>109135898how exactly? this seems like incredible proof that all of this AI shit is working. and the US is in the lead, not china. this regulatory capture (>>109137601) means the US is going to stay in the lead.
Works for me
Jeffrey Epstein's other close personal friend and island buddy, Howard Nutlick, has been involved in banning both models.GPT 5.6 has been ready to go for WEEKS.
>>109137258I mean, they're probably better than the old ones but they can't afford to provide them to the masses due to increases in compute requirements
>>109137469>Let me guess, they are going to get away with it?Well, duh, the US public is too timid to ever dare go up against the tyrants that control them.
Why the FUCK do AI people talk like this?https://x.com/theojaffee/status/2070331417851310283>16 months of service>beautiful mind>big model smell>I'll miss it very much
>>109135387oh come on, it can't be that amazing. The gov just trying to hype it up
>>109135757why do they all have the same nose then
Why are they holding back technological innovation? Don't we want AGI?
>>109137620You can't stay in the lead by staggering releases. You can't defend yourself if the people who need those models can't have it.
>>109137620The US will restrict their general population from using it. China will engage in industrial espionage, steal the models that America invests in training, and then let people use them at will. This will lead to increasing hysteria from American jewry.
>>109137140>I wonder if they'll start war with China over the next year.They couldn't beat Iran and keep the pool open, what make you think the orange retard has a chance against GOD Xi-sama?
>>109135387GPT fucking sucks like Grok now. Chinese models are the only decent ones left.
>>109138515Industrial espionage like...using the model and training it like that. This is not something anyone gives a shit about anymore. We all know this bullshit is just a way to make the absloute black hole of funds and rampant fraud of the "AI" companies into a state issue so they can get a bailout. We also know that the genie is out of the bag and local LLM and old bitcoin mining rigs can do 99% of everything the average person will need. This is theatre. A way to convince you that it's okay when the US government bails them out to the tune of trillions of dollars.
so whats the difference between this country and china now?
>>109138762The US labs made enough money from serving each of their models that they paid for their training costs of those respective models. China subsidizes all its models as a form of economic warfare against the US. You're welcome to say that's based of China, but don't be surprised when China catches up to the frontier of AI development and suddenly their best models are only available to CCP approved companies and the open-weight export-only models are years behind and trained to insert backdoors into your code.
>>109135387It is staggering how fake, gay, and completely lame the tech industry has become. As though 1980s IBM shunted the timestream and took back control.
>>109135387i dont carei was not going to use them anyway
>>109139125only thing thats going to save us is open source.
Chinese models mog both tho
andrew curran is a convicted sex criminal with a diaper fetish
>>109139270so is donald trump tbf
>>109139293That's actually incorrect. Trump lost a civil case, not a criminal one.
Guess I'll go with Mistral.
>>109135387>"nuh uh only we can use good shit">trillion dollar valuations lower by 80% since the market goes from "a few % of most of world’s economy" to "a few % of murrican economy">the rest of the world switches to providing all the free training data to uncucked chinkmodels>this causes chinkmodels to close the gap by mid-2027this seems really questionable as far as fuck-fuck games go
Funny, Dario and Sam's constant doom trolling was meant to earn them a $1 trillion evaluation, but instead they taught their customers that they should use other models instead
>>109136687MUH FABLEMUH MYTHOSThey are LLM chatbots you stupid fuck, do you actually believe any of the absurd claims coming out of Dario?
>>109135387>you can't own hardware goy>we need hardware for AI>actually, you can't use AI either>only the (((selected))) may use AI>no, you can't work or reproduce either>we've got AI to do that
>>109135598american golems have no mouth and must screamtheir only change to do the right thing was to embrace Adolf Hitler
>>109139579amerigolems had their moment in the capitol riots. but they ended up doing it for israel kek
>>109135412Sneed more, slopper.
>>109139421It's not funny it's sad. Dario genuinely believed that AI can be dangerous and was willing to put some amount of safety before some amount of money. Sam wasn't willing to do that, but everyone nose why that is.
look at the miggers itt doing gymnastics to justify why it's the fault of the people releasing models why the models are being withheld from the publicfable was already outyour pathetic migger brains may break if you have to admit that you've been psyopped for the last 10 years but you're just embarrassing yourselves
>>109135397is anyone outside the US even buying these models?
>>109140249>Anthropic’s recent Economic Index report found that nearly 80 percent of consumer Claude usage comes from outside the United States, with per-capita usage in countries like South Korea, Australia, and Singapore outpacing that of America’s.that's usage not revenue but it's likely still more than 50% non US.
>>109135397>They spent the last 18 months playing up the national security implications of AI chatbots and now the US government took the bait.So you're saying that a frontier lab is so powerful it has managed to trick the US government into changing its national security policy in a way that harms the US?That sounds like a reason why these labs SHOULD be under more government control, so they don't harm the country even more.If Trump agreed that next year a country of geniuses in a data center would suddenly appear within the US borders and not be under his control, he'd threaten to invade it like Greenland or Cuba.That's almost what happened after the Pentagon declared them a supply chain risk.
>>109140214As a non-American I think your president did the right thing in limiting AI.
>>109138609Claude is absolutely wonderful as a former ChatGPT loyalist.
>>109135387Yeah because the Chinese will just distill it. Welcome to the future we luddites warned about, no more models for the public but you still need to play your vidya on jenson's GPU farm
>>109139305This. He's a liable civil sexual assaulter, not a convicted rapist. And his diaper wearing is likely not fetishistic, but strictly functional.
>>109135387I’m still using got 5.6 and Mythos but then again I’m not a goyim with just a slightly bigger paycheck than normal
>>109140623He's as straight as a New York property developer with a silver spoon from daddy
>>109135387I feel like China is going to win the AI war now.
>>109135387Americans are allowed to, but because Dario is a leftist extremist he doesn't want to ban non-Americans, he instead banned for all Americans
>>109140743Claude has been more open and free in terms of thought for me than ChatGPT-5 series has been.
>>109140700>Sam: We can't let you have GPT 5.6 (which is totally ready, by the way) because Trump says it's too dangerous.>Susan: Awww, you're so sweet.>----------------------------------->Dario: We can't let you have Mythos because we want companies and open source projects to patch their zero days first.>Susan: Hello, Twitter?! Anthropic has run out of compute!
>>109140769You let your computer think for you if that truly freedom?
>>109140149Yeah... No. Dario is undoubtedly intelligent considering all his credentials, which is why I have a hard time believing he actually believes all the absurd claims he makes
>>109140743The only way Dario can check the citizenship status of every Claude user is to integrate with a federal citizenship database/API run exclusively by Palantir. If he doesn't do that then Trump can execute him for treason and The Reptile doesn't get his quid pro quo for helping Trump win the election. Trump is delaying the availability of that API in order to do maximum financial damage to Anthropic. Trump knows he can get away with corruption like this by scaring people into believing that the other side are communists.
>>109140798I don't, I have it interact with my original thoughts and assist in their development the way another person would. In that aspect, Claude is more like an expert and ChatGPT more like a Redditor.
it's actually crazy, if they start getekeeping software creation again right after they just gave us all access I'm going to be fucking pissedhow are more people not up in arms about this
I give it 6 months until theres an open source model that might both mythos and GPT 5.6, sure anthropic and OpenAI will probably have better models themselves by then but the reality is AI cannot be stopped.
>>109135387its a chatbot it doesnt have 'security concerns'he's just bullying them to get leverage in some deal. they tried the 'ooh my chatbot is dangerous' marketing meme which worked on no one and the us gov made them swallow their own words by pretending to treat it like they do in their marketing.>ooh if this chatbot is so good its 'dangerous' that means you cant sell it right
>>109140919remember 2020 and how we all did nothing about global lockdowns and nonsensical mandates???nothing. ever. happens.
hey /g/ I don't know where to ask this but what is a lightcone and why are techbros freaking out over it
>>109136787Inference is profitable, R&D is not. Reduce the number of users and now the only part of the company that is profitable gets hamstrung.
>>109141006“everything in my lightcone” is everything I could possibly affect given that the speed of light is a universal constantstuff outside my lightcone is stuff that’s further than a light-year away from me when I was one year old, or two light-years away when I was two years old, etc.
never seen a better reason to switch to chinese models KWAB
I'm perfectly satisfied with gpt 5.5. I am an extreme power user and have been paying from GPT since the day it was possible to do so, so I think I'll be one of the first approved for 5.6. I do wish I could host 5.5 locally but that just isnt possible on any hardware I could reasonably own. Might have to treat this level of compute at the same level expense as a home generator or a car? ~26k for compute in my basement to run 5.5 seems fair for the benefit I get. But I have a feeling the actual price for this would exceed 80k
>>109139444They aren't magic but everyone is already using Opus and Codex for reverse engineering and cyber security. I know those models work, maybe not better than the top human experts, but imagine every retard now being able to penetrate medium security systems. Fable was a bit better than Opus the 3 days I used it, so it could absolutely become disruptive.
>>109140988>they tried the 'ooh my chatbot is dangerous' marketing meme which worked on no onenot even the NSA?
>>109135387>>109135397>Both AI companies are going IPO later this year>suddenly, suddenly, their models are "too powerful" and dangerous to anyone outside of the US and even inI'm sorry. I know that autistic people have trouble reading between the lines but this is like right in your face kind of shit. It's very obvious that Trump is doing business favors for them to make their companies look more powerful than they really are all of a sudden.
>>109140883All you have to do is check bank records, is that commies and thirdies fear?
>>109135387All vibe sloppers are crying now because their life and wallet is in shambles höhöhöhöhöhöhöhö
>>109141184yesnsa has people who know what chatbots are
>>109140988They're just worried about the patch cycle breaking. "Open" models from the big two are a lot better at finding security vulnerabilities than most people realize, Mythos et al isn't a big upgrade in terms of raw bug finding. The existing models are just bad at validating findings. Main improvement from Mythos appears to be actually validating bugs it finds. But that itself has security implications because now instead of needing a human to validate the outputs an attacker can autonomously patch diff the entire internet and pwn everyone who didn't patch instantly. So big industry players are afraid of having to make broad changes to the ways they roll out software updates. If they were just afraid of Mythos et al finding crazy exotic exploit chains they would be happy with simply having advanced access to the models so that they could find and patch the bugs before they ship the software/updates. But the fable jailbreak fiasco shows that that's not the thing they're actually afraid of, since simply having access to the models yourself solves that problem. Eventually they will realize that in a post-mythos world you cannot just ship a security patch and call it a day, since most large organizations can't instantly patch their systems without testing the update first and making sure it doesn't break stuff. They'll need to first ship mitigation and indicators of compromise and make sure that their software is designed to be resilient enough that applying these mitigations with minimal testing doesn't break everything. But that would take actual effort and crying wolf to the government is easy. So it'll probably take China releasing a similarly capable model and a bunch of people getting owned after a major security patch for anything to actually change.
>>109141008kind of moot given users are expecting iteration every month in this climate. i'd say the r&d is more of an operating cost at this point.
>get immediate access to the new model>it still cant solve your problems because it's a retard>it still cant do anything cool because its lobotomized like all commercial AIvery cool
>>109141141>Fable was a bit better than Opus the 3 days I used it, so it could absolutely become disruptive.NTA buy fable was monumentally better. Opus can follow moderate level shit but once you have multiple threads, asynchronous systems or subtle race conditions it falls apart.Fable can go from a brain let prompt not much better than "make it work" and nail all that shit in like 30 minutes, worst case.I had Fable knock out a 3 month subtle bug in a matter of 30 minutes that was like 4 timing systems intertwined.
>>109135387and this is why we will get to the point where these models are locked away even though we helped train them, and we will have to run our own local modelslets hope chinese continue to beat the western AI
>>109141184there is an extremely massive difference between breaking in via handing mythos their source code, and mythos actually being able to hack
>>109141303Anthropic gating off frontier LLM research (among other things) for Fable when it was out makes me suspect that more than simply raw bug finding it's knowledge about specific topics and subjects which the smaller open models do not internally have that is scaring the US gov/safety cultists.
>>109142081more like the fact that all LLMs speak russian and chinese alongside english, some chink posted the NSA master password on the darknet and antropics crawler accidentally scraped it and put it in the model
>>109142011yes, but the first is spooky enough
>>109142107security through obscurity is literally the first line of defense for all infosecfuck off all retards
>>109141424No, they aren't. They are expecting the product to get better every time it *does* update, not worse, as has been the case for 12 months now.
>>109141094"People" talk all the time about how chink models are "superior", but never about what they've actually done with them.
>>109141105What exactly is good about 5.5 in your experience?
>>109142239I use DeepSeek V4 to do mundane coding tasks because it works for almost everything. If it can't generate a good solution, I try Claude Opus. Do """people""" give you detailed diaries and video essays about what they achieved with frontier models from American companies?
>>109141184This is a ChatGPT thread.
>>109141956Very interesting. Anything more to say on this?
>>109142257>Do """people""" give you detailed diaries and video essays about what they achieved with frontier models from American companies?I'm one of those people observing my own firsthand experiences, which is how I know that anyone who puts them down to prop up chinkshit is bullshitting.
>>109142257I actually did elsewhere
>>109142231who's using anything other than the latest models?
>>1091423444 months ago plenty of people were still using 4o. Nowadays most people literally don't have a choice.I tried o3 since it got made available this year, and I thought it was trash.
>>109142265Nothing specific but ask away
>>109142424What did you observe as being the genius of Fable, and how do you think it works that way?
>>109142454The interesting thing is Fable didn't spell out reasoning like other models. It just worked. Lots of tool calls. Almost as if it deliberately was told to suppress explanation of its reasoning.I do reverse engineering of games. With Opus I have to step by step, articulate an issue, give it clear reproduction, check in frequently. Pivot it, roll back to safe commits often. Doing even a single game is a nightmare but I can make it happen.With Fable it really is "I have my setup, it boots but stalls, make it work end to end" and by fucking God it actually managed to more or less identify all current and future pitfalls and more or less one shot them.I'm eagerly awaiting it's return.
>>109135482What race is that thing and why it looks like he was cursed by God and lacks soul?
>>109140214>The government needs to regulate AI more>NOOOOO NOT LIKE THAT WE NEED TO BEAT CHINAyou faggots have no consistency
>>109141184There's this guy and there's David Grusch who claims that interdimension aliens are regularly crashing UFOs but only in the US.
>>109135783ack-tropic and vibers around it a conmen and leftists
>>109135387Literally, they have a group of jews evaluating it for sufficient philosemitism before the goyim are allowed to ask it questions.
I wonder if it really is getting to the point they can't stop it from naming them.
>>109142239they don't get blocked by their government, for starters
>>109141424They have been yes, but eventually we'll reach some kind of Pareto optimized state where cost makes even better models unpalatable to the majority of users. So it's more of a scaling cost than an operational expense, and every private investor expects their investment to put their money towards scaling so that's a totally acceptable deployment of capital in their eyes. The current SOTA models are already very useful, I don't think it would be a problem if we only got 1 or two more major improvements. 5.6 and Mythos are probably very close to that Pareto optimum though.
>>109141956If you lead Opus to water it will drink whereas Fable seems to have a dowsing rod. I have found all of those bugs and more with Opus and GPT but it requires a lot of structure built around the model that Fable doesn't need.
>>109142801Wow. This is how far modern 4chan has sunk, claiming the CCP isn't obstructionist.
>>109142589> told to suppress explanation of its reasoningit literally was, they said this publicly. There's a dataset of Mythos preview thinking traces on HF and they are each 200k+ tokens.
>>109142860I still have access to the Chinese models, don't I? not my fault America decided to go full retard the last year or so, keep seething jew
>>109142248I can use it for work and generate scripts and programs to automate any task I need, just using codex. From running applied test cases or system calibration scripts, converting my ideas into programs for testing that can be minimally viable and sent to the software team, writing white papers, and even stupid shit like interacting with excel files and processing tons of data for bullshit coloring book tasks I don't want to do. With 4o it genuinely made mistakes and it was really laziness driving me to use it, so I can bullshit on my phone and debug things later. Now it can actually complete tasks
>>109142985you could do all of that with GPT-OSS 20B and a 16gb GPU from 2020do you have any idea how wasteful you are?
>>109142606Here's an analogy for you, if you can understand analogies.>Dario: Hey, city council, I think we should have a fire department for the city, in case a fire burns down one of our houses.>City councillor: Okay, so we're going to waterboard you until you give us the deed to your house. That way you don't have to worry your house any more. Fire departments and waterboarding both use water to achieve their goals, so this must be what you wanted.Just because something is ironic and cruel doesn't make it good policy.
>>109143165disingenuous faggot
>>109142598God's Chosen™
>>109135387i'm voting dem at this point fuck these guys
>>109143188>thinking team donkey wouldn’t do this tooso you’re a gay race communist over all else, got it
>>109139579American media convinced their goyim this orange pedophile slumlord is somehow equal to Hitler. Laughable
>>109142990okay nigger, go ahead and do it then. I've been using GPT since release, I work with local models. I remember the first "good" deepseek codex models, they sucked.
>>109143165
>>109143304I'm not defending the company. I'm saying Sam should be waterboarded too.
>>109135506By making crap models? China is 5 years away from mythos/5.6 AI
>>109135662This is terrible for their potential IPOTrump just wants to use mythos/5.6 as a bargaining chip and miggers will clap
>>109135757You need to be invited into the tribe. You can’t just convert. Or rather you can but you will always be a goy to them
>>109142860he didn’t say that, schizo.
>>109143226I already do, all you need is the agent harness if youre so successful with your cloud API, why not just ask it to set it for you? it's literally free to try unless youre using a chromebook
>>109143494Okay, how? Describe EXACTLY what you do.
>>109142853That's my take exactly. I put a lot of structure into my ecosystems but even the more complex problems can elude it.
>>109139990as you wanted >>109123836
>>109140700People keep saying this but nothing that's happened so far has prevented these companies from continuing to develop models like Mythos/Fable and GPT 5.6. The only thing they can't do without government acceptance is let everyone use them. The US is figuring out how to handle AI models that are a legitimate step change. On top of that, it seems like the govt themselves now wants to make it harder for China to distill (which they are definitely doing).If anything, only allowing certain companies and/or groups to use this step change tech means that the US will probably pull further ahead due to basically no Chinese distillation. On the other hand, we're getting a preview for what the future will look like: only big tech and govt getting to use the truly frontier and closest to AGI models. I do feel like we are barreling towards a legitimate cyberpunk dystopia, but damn if it isn't an insanely interesting and entertaining ride.
>>109137097>implying china won't ban non-chinese access in the future
>>109137733>Yid cow skyHow do they have such comical names
>>109139081>Le Chuyna>Le subsidise commie>Le US>Le free market subsidies:)
Kayfabe.
>>109144966You really can't stop distillation meme without banning api usage, why do you think musk bought cursor?or google offered claude for basically free for months in their ide?You can distil whatever the fuck you want as long as that market of pay to integrate our ai in your app remains, hell i am very suspicious of the whole ai battle meme were they let people use different ais and compare results it's like giving something to the public for free so they select best result thus vote for you on good material for ai training.The us gov has an advantage because anthropic and openAI, google and others failed so far to do something revolutionary, but it's not that much of an advantage and frankly i do not see how the fuck can you keep such "monopoly" for long, it's like a temporal advantage like the steam machine and trains was for britain eventually everyone will get the same economies of scale, the game is in getting first faster and with good regulatory system not to prevent others from getting it because you really can't desu.Also i am sure open ai models will be opus 4.8 level soon, at some point some smart ass will start to autodistilate each other kid of an ai vs ai battle and create a super open ai model that will be "filtered" with the best of open models and will have a perhaps better base than the censored top models.Frankly i think reason they are scared is because most important systems are probably still using cobol or shit like that and smart asses will start to break into things, not to mention the retarded cuck your customer system they implemented to cuck crypto in the entire banking system can probably be broken with ai larpers using fake video, images etc to the point the banking system will have to return to fucking in person transfers or fax orders.That said ai is a tool the solution to this problems is better security like passkeys etc but illiterate boomers are insane.
>>109135387>AI is largely unpopular with the masses.>2028 is rapidly creeping towards us.>Every political group of America is becoming more and more anti-AI just to appeal to the voters.Woah, wow, what a surprise! While Trump himself will not become US president again without some major changes in law, obviously he will setup things for whoever will get his endorsement.Anti-AI politics are an easy way to get support of the masses nowadays.
>>109135387Damn that sucks but I can get 90% of the functionality from deepsneed (its free) and glm 5.2 (more expensive than usual for chinkmodels but its opus 4.5 tier) so I guess its chinks that get free training data from me now>cheap solar panels>open weights ai sloppa>soon cheap batteriesGod bless china, unironically
>>109135387CANNOT REDEEM THE CHATGEEPEETEE 5 DOT 6 LUNAORANGE BASTARDBITCH BLOODY
You're mostly right to be concerned. The real question here is whether people need it. You don't.
>>109135387Well i guess china or jewpan will reach AGI first then
Fable was so fucking good and I absolutely need to have it again.
You don’t need more than llama 3, chuddy
>>109146217Sure you need it, there is a reason why works of art like 2010s games like fallout new vegas, mass effect, skyrim and others stopped happening.As inflation rise so did cost of long term projects so the entire economy become about rent seeking not investing in long term projects.AI actually reduces cost of investing in long term projects a lot you will even be able to make movies with ai only in some years.Also it's irrelevant what politicians say open models are 7 months behind gpt 5.5 and opus 4.8 now.>>109146275It will probably be china specially if they go ahead with the whole anti litography export shit, the chinks will then have a reason to do a manhattan tier project of open AI models simply because you can distilate each other and having them released worldwide for free means a ton of gpu demand from autists running better models than anthropic or open ai models (if the usa bans them), which means china will then be able to smuggle chips from commercial markets unless usa literally ban gpus.Point being this stupid arms race can't be contained anymore, you can slow it down but whoever contains it losses as the best training material is from the 2005-2015 internet era which will be disappearing as time goes along and people simply ask things to AIs.So it's a now or never and both usa and china are positioned to dominate the ecosystem, if you think china releasing open models is weird just imagine what would happen if they are pushed away from chip manufacturing and need to start smuggling gpus.They will start financing the release of open models simply to increase the consumer market demand worldwide.
>>109135387cant wait for chink alternative to it lmao
It's ironic that China still remembers the lessons of the Opium Wars but the West have forgotten. Businesses will happily fire workers leaving us all deskilled and dependent on UBI, then China will rugpull access to future models and it'll take us a generation to catch back up to the frontier.
>>109137258>"true money"silver is a rock with no inherent valueactually just as much as fiat
>>109147819>>109147973I don't see China treating their models any better if they're truly becoming a threat to national security. What will happen is consumer models will stagnate and all future innovation will be for military use. A 1.5 trillion dollar military budget will be peanuts in the next couple years
>>109135387>companies are in massive debts>let's not allow them release their productsbased trump accelerating collapse. ram will be finally cheap.
>>109148016micron has contracts for at least 20% of their supply until 2030, can assume similar stuff for the other two companies, really doesn't seem like prices will recover soon, certainly not to mid-2025 levels.
>>109147993>silver>no inherent value>He says, post on the internet, on a computer
>>109148006China will allow state owned companies to use their latest closed weight models, with guardrails to prevent the models being used for dangerous or subversive activities. Every request will be logged and associated with a Chinese citizen's national ID number. Some Chinese model hosting companies will be allowed to sell tokens to Western customers, if the use case is approved by the CCP and it increases economic dependence on China. Allowed use cases might be Western journalists writing news articles (the models will use approved CCP terminology) and Western lawyers writing business contracts (the models will track Western supply chains to look for further weaknesses).
>>109148141You CANNOT be this retarded.
>>109148141you got me interested in how much of silver's production is used for industry rather than investment or vanity>Industrial applications consumed 680 million ounces of silver in 2024, representing 59% of total global demand. This marks a dramatic shift from just a decade ago when industrial uses accounted for roughly half that share.
>>109148113Contracts can be changed. AI fags need data centers for their large models, but if those models aren’t allowed to be used except for some military shit, which are nowhere close to consumer-level usage, It will be very hard to justify having massive data centers sitting there doing nothing while being trillions of dollars in debt.Also, governmental bans are quite painful to get rid of. shit ton of bureaucracy and politics, which take ages.Mythos will be banned for at least a year for regular consumers and businesses. That means hundreds of billions in lost revenue for Anthropic.
>>109148205>That means hundreds of billions in lost revenue for Anthropic."The cruelty is the point."
>>109148205>It will be very hard to justify having massive data centers sitting there doing nothing while being trillions of dollars in debt.Thats why goverments are desperate for digital ID and age verification they are trying to legislate into being a use for AI so your tax bucks can pay for the tech bros massive massive investments into data centers.
>>109148205>That means hundreds of billions in lost revenue for Anthropic.for OAI as well, because they aren't allowed to release gpt 5.6.and glm5.2 is super close to gpt-5.5. they are super fucked.
>>109147993>t. vampire
>>109147993>werewolf claws typed this post
>>109135387Creating false scarcity to drive up demand, and the cattle will lap it up like always. Just ask Dawkins
>>109135387>"our new SUPER ADVANCED HYPER ULTRA AI models are TOO DANGEROUS, it can DESTROY HUMANITY, pls gib us more moneys">get regulated coz tech challenged regulators believed that>banned models are mid upgrades to old shit, 5% better trustmebro benchmarks, 10% more hallucinations
>>109148649>5% better trustmebro benchmarksso you would predict that the number of publicly reported software vulnerabilities found in the past couple of months would be similar to the amounts found in the equivalent periods of previous years, right?https://epoch.ai/data/cve?view=graph
>>109148141>Buy a rock whose main uses are last-gen solar panels and reselling it to other rock collectorsActual sub-fiat status
>>109148833Maybe the jeets are just really sloppy and now they have a fancy auto complete to exponentially ship jeetslop did you ever consider the bad software getting shoveled out might come with extra security vulns?
>>109148229Digital ID doesn't require massive data centers full of GPUs.
>>109148871these vulnerabilities have been in the code for years or decades. if the latest models are finding cves that previous models didn't then that means the models are improving.
>"Guys! These new models are so advanced! They found lots of exploits on gov slopware!">But even Gemini can find those>NO! It's too advanced! No one can use it!
>>109148898>>But even Gemini can find thoseDon't give those retards any more ideas.
>>109148898>But even Gemini can find thosehttps://www.aisi.gov.uk/blog/how-fast-is-autonomous-ai-cyber-capability-advancingIn AISI’s latest testing, the newer Mythos Preview checkpoint completed both our cyber ranges, solving the range “The Last Ones” in 6 of 10 attempts and the previously unsolved “Cooling Tower” in 3 of 10 attempts. This was the first time that a model completed the second of our two cyber ranges.
>>109135387Free market btw.
>>109148966We haven't had a free market in decades. The government and federal reserve act as central planners, funneling capital where they choose, picking winners and loser.
>>109148976Wtf!? I hate America now!
>>109148987It's not just America.
>>109135387Jokes on him because I RAN OUT OF TOKEN ANYWAY
>>109135387lmao I'll just use Chinese AI's
>>109137251My guess is that faggot Altman was talking into his ear to sabotage Anthropic and it backfired spectacularly.
>>109149944More likely Trump got this idea from Jensen who was mad that Dario called him out for selling GPUs to the CCP. Jensen doesn't care if OpenAI gets hurt in the crossfire.
Where the fuck is picrel in all of this? Why aren't they stepping up to help balance the power?
>>109150339google deemind london is hemorrhaging talent because everything's moving to silicon valleythe uk government is going all in by investing - 1 whole billion pounds lolmistral is the closest thing to a frontier lab europe has and it's at least a year or more behind the frontiereurope doesn't have the money, the will, or the talent (they all went to america because that's where all thet money is)they're going to spend the next decade begging america to not cut them off
>>109150339what power?what you’re showing me is a mid-tier mostly muslim part of the world like indonesia+malaysia the only real countries anymore are the US and China
>>109150339lol
>>109135387Doesn't matter state funded hackers that target grid will just use something based on open-source models.>>109135506China won by not being multi-cultural hellhole, they have more time to develop whatever than any western country. Shartmerica will become Brazil 2.0 and Evropa will be North Africa tier.>>109135822He probably sturges to use macbook, makes sense.
>>109150553>the uk government is going all in by investing - 1 whole billion pounds lolThe major labs are spending hundreds of billions of dollars on capex and talent. Even if the UK government had that much available, and even if the general public wanted the government to spend that on AI rather than improving healthcare and defence, what could the government realistically do? There's no way that a government owned project will be more efficient than a privately run one.>>109150634>mostly muslim part of the world21160000 Muslims out of a total EU population of 446470000 in 2020 (excluding Cyprus). That's 4.7%. The US figure is 1.2% for comparison. If you think that less than 5% still counts as a majority then right wing propaganda has broken your brain. Even if the EU managed to get rid of all its Muslims, it still wouldn't be able to compete with the major US labs.
>>109151085compare person-hours in mosque vs. person-hours in churchI stand by my statement>Even if the EU managed to get rid of all its Muslims, it still wouldn't be able to compete with the major US labs.100% agreed
>>109151221What is the equivalent building for an atheist? A house? If you mean the religious observance of the EU is mostly Muslim then you might have a point, but religious observance in the EU is itself a small minority of what citizens spend their time on. In fact I'd be interested to know how many EU person-hours are spent in workplaces for online and digital technology businesses.
>>109150339Europe is irrelevant and part of the third world.
>>109152074The US has poached all the talent and bought up all the European startups, because of the way capital accumulates. Fortunately Europe still has a couple of cards up its sleeves. One is the fact that they control ASML and some of the supply chain which ASML relies on. The other is that the UK and France are nuclear powers. (So is Russia, which is arguably part of Europe, but this argument still works even if Russia and the US join some kind of alliance.)I know this sounds crazy, but the UK and France need to run a tabletop exercise where a US lab achieves Recursive Self-Improvement, leading to automation of most of its economy, and 20% industrial and military growth every year. They will realize that under such a scenario, the US could decide to revoke foreign access to its AGI or ASI model, causing Europe to enter a recession and eventually become vassal states to either the US or China.The only way to prevent this is to threaten military strikes against US electricity grids, fully acknowledging that this will escalate to a nuclear exchange that both sides will lose. If the US don't want to risk calling their bluff, they will have to make the concession of internationalizing their AI labs and supply chain, giving European countries access to some mandated minimum share of GPU production, and letting European inspectors review the hardware-enforced guardrails of the models that the US deploys. The two risks that Europe needs to mitigate are having their access to hardware cut off, and the US using its models to find a way to neutralize Europe's nuclear deterrence.
>>109150339No place in Europe is a super power. Russia roleplays as one. It's literally just us in America and China.
>>109152305>they control ASMLstopped reading right there
>>109135662He has to keep the show on the road until September / October for the IPOs and this is one way to do it
>>109152349>stopped reading right thereI may have over simplified the hard power options and limitations to what the Netherlands could do given its current level of influence over ASML, but if you're not prepared to read the rest of the argument or state what detail you think I'm missing, I'm just going to assume that your objection is invalid. To be clear, when I talk about Europe, I don't mean the EU as a supranational government, especially as the UK isn't a member of that union any more. I mean that region of the world and the nations within it, working either unilaterally or multilaterally.
>>109149027Why not use an open-source local model?
>>109135440>10 years of unregulated AI
>>109151085>That's 4.7%I don’t believe that anymore.If I go outside in Germany I feel like at least 1 in 3 people is Muslim or foreigner
The technology was stagnating already, and adoption was high enough. Time to IPO, stages regulatory capture, and enshittify the product.
>>109153036Facts don't care about your feelings, cuck
>>109153036Germany might be above the EU average in terms of Muslim population. Also, it's difficult to tell if someone is Muslim just by looking at them. Even if they are from a Muslim majority country, that doesn't mean they still identify as Muslim. If someone is Chinese, that would count towards your 1 in 3 number, but probably not to the 4.7% number. One other factor is that you say go outside, where you are more likely to see beggars and muggers, whereas if you were visiting a retirement home you would more likely see old white people who don't get out as much. Ask an LLM if these factors are enough to explain the discrepancy, and you might also find out how much political bias the model has.
>>109153659In Germany it's very dependent on where you live.Something like Kreuzberg and rural Meck--Pomm are completely different worlds.
Time for China to shine?
>>109135387>Do nothing>Win
AI labs finally got what they wanted.
>>109140700>gayJ weirdo who scammed pretty much everyone, molested his sister and murdered his own employee>literal J$ island VIP, also part of "the club"These are the people making decisions for the USA now.This is actually the worst timeline. WTF happened? At least Chink models are open weight, which is endearing me to them.
>>109143165Yes, except Sam and Dario both knew what the government would do. They want the moat. Regulating AI is the infinite money glitch. They already bought the hardware market out, now they are implementing legal hurdles for everyone who wants to compete. So much for capitalism.
>>109143386Based off what? You can't use it to see that it's significantly better. They basically just said "I have a way cooler toy than you. What? No, of course you can't see it!", and you believed them.
the level of psychosis or shilling about fable and crap is unreal on this board lol
>>109155599I hate what tech has become
do you think donald trump understands the difference between a monitor and a computer?
>>109155935"Everything's computer!"
>>109155374>They want the moat.What moat? How does delaying their model releases help them? The regulations don't apply to Chinese open weight models, or to small startups. It only hurts the biggest labs with the deepest pockets who now have to pay bribes to Trump to get permission to do what they were previously allowed to do without paying those bribes. Businesses want clearly written, predictable regulations, otherwise they lose investor confidence.
>>109156458you're supposed to just say buzzwords like regulatory capture and moat, not actually understand what they implyhold on, let me show youthe regulatory capture will help inflate the bubble allowing investor to take everyone's pensions and invest them into the datacenters via SPV's, that will then emit infrasound frequencies that will drive most of the population to suicide and the rest will die due to lack of water
>>109156543As I'm someone with no understanding of how investment works, you convinced me just by saying the word bubble. Well played.
>>109155935What's a computer?
>>109135387if the Biden administration did this and the Anthropic thing, people would be creaming their pants praising him and all the major news outlets would be posting positive news article about this decision
>>109156669yeah but that didn't happen
>>109140883Yeah, Anthropic is very clearly regarded as an enemy of the state which indicates there wasn't sufficient bribing towards trump or the other open criminals in office. The surprising thing about this story is that it's affecting OpenAI, but it could be a reward for continued loyalty from musk since grok he just acquired Cursor and needs to catch up. The thing with current business and government in the US is you need to regard them all as extremely corrupt criminals having turf wars. People need to apply what they know of banana republics and cyber dystopias to better understand our current world.
>>109141198Are you trying to say Donald Trump, the world's most honest man, is manipulating global markets to further the profits of a select few of his friends and family? I'm shocked you'd even have the balls to type such lies, let alone press the post button. This troon anime image lush farm of an image board deserves better than your lies anon. For shame.
>>109135387Donald Trump will lose the next election for the retarded reason of RAM being expensive.If people can't buy the next new thing, they will vote for the fucking dems.
>>109156669>if the Biden administration did this... it wouldn't be an unannounced threatening phone call, it would be an Executive Order saying that independent technical experts would assess the cyber attack capability of all models that were trained on more than a billion dollars of compute. Trump did issue an EO, but it said that the assessment was voluntary and that there was no preclearance requirement.>Nothing in this section shall be construed to authorize the creation of a mandatory governmental licensing, preclearance, or permitting requirement for the development, publication, release, or distribution of new AI models, including frontier models.
>>109156907...? he can't run for a third termand he would not downgrade himself to being the vp or a senator
>>109156863>Yeah, Anthropic is very clearly regarded as an enemy of the stateHegseth wanted Claude to be able to do autonomous second-strike capability in case we were under massive attackDario’s response was “oh, in that case, call us” (and basically I’ll decide if the US actually has a nuclear deterrent or not)Hegseth was 100% right to treat this as a coup attempt and regard Dario as an enemy of the state
>>109156960>Hegseth wanted Claude to be able to do autonomous second-strike capability in case we were under massive attackNo, that's just the excuse they came up with after the fact. Anthropic knew that militaries want to outsource their target selection process to AIs that are not capable of doing that reliably (see the Iranian school bombing for example). Dario worried that people would view his company as morally responsible if his AI picked the wrong target (for example, if it made US drones attack US warfighters because Claude is legally blind), so he wanted the model to not be used for that use case, even though he knew that other models would be used instead.>Dario’s response was “oh, in that case, call us” (and basically I’ll decide if the US actually has a nuclear deterrent or not)Anthropic never said "call us". That's a lie that Hegseth invented. The contract negotiation was about the specific types of systems that the Pentagon could integrate Claude into. As the creators of Claude, Anthropic knew better than the Pentagon what the limits of their AI are. No one should want unreliable AIs deciding whether to launch nuclear missiles, and at which targets.In a perfect world, the AI companies would be able to accurately characterize all the failure modes and reliability metrics of their systems, and then the Pentagon would be able to decide whether the risks of accidental launch or friendly fire were too high. We don't live in a perfect world, though, and the AI companies don't know what they've built and the administration thinks that a weapon being criticized by ethicists makes it cooler and more desirable.
>>109157046what a fucking joke. Anthropic has already been used to kill palestinian children by israel
>>109157465The people who care about Palestinian children are already opposed to anyone giving any money to any AI company, not just Anthropic. Also, most people don't know about Anthropic's connection to Israel because the mainstream media doesn't discuss the details of Israel's crimes, for some reason.
>>109155935Say what you want about drumpf, he is definitely a power user if you look at how many sloppas he post each week.
>>109135387The models are soo powerful that the Goverment will have to buy the companies at premium price with taxpayer money.
>>109157465That might have been Grok.https://www.independent.co.uk/news/world/americas/us-politics/elon-musk-grok-ai-iran-missiles-pentagon-b2997321.htmlI don't think we'll ever know, which is by design. Blaming an LLM for war crimes is just too convenient.
>>109135822it's not orange man boomercession, it's intel community briefings
>>109158077>buy the companies at premium price with taxpayer moneywhy would the government do that when they can just extort bribes from the companies and leave them with the toxic assets?
>>109158621leftists can never ever say "netanyahu is a war criminal". the powerful jews of leftism in every country prohibit that.
>>109158696>https://www.sanders.senate.gov/press-releases/news-sanders-introduces-legislation-to-create-7-trillion-ai-sovereign-wealth-fund/Because its being setup to bail out ai companies.
>>109159016The pol rule is so reliable and refreshes the mind, you find a pol thing, just nope right out. For example, you know that pol-type guy who reviews tech and is always getting recommended, and who gave the apple goggles a rave review?It's liberating power to know these things, and thereby know safer paths.
>>109159016it's difficult to get the senate to agree on anything but even harder to get them to agree with an idea introduced by bernie. taxing the shares that someone already owns is a wealth tax so that would need a constitutional amendment too.
>>109135387Isn't Donald Trump too fat and old to speak himself anymore?
>>109159106(laugh track)Did you hear about druuuuumfffff(laugh track)I was at walmart the other day just kidding I'm not a pedo trump voter(laugh track)seriously why is he so orannnge(laugh track)
>>109142413Man I miss 4o
>>109135387>we must do whatever we can to win the AI race and ban you from using them
>>109135387"For security concerns" just say how it is, it's for total information control. Gpt 5.6 is no different from the previous gpt5 versions other than some optimizations. They just don't want people to have an automated researching tools.
>>109135387I would rather not using them to begin with, specially with personal stuff
>>109159912There is plenty of information that is freely available and doesn't require automated researching tools to find but people don't know any of it because they prefer to be ignorant. Political change is not bottlenecked on the availability of better tools.
>>109159958>he thinks the printing press didn't cause political change.
>>109159993Information is like water. In the past, people were dying of thirst. Nowadays they are drowning, but they prefer the taste of poison.
>>109160109Its not about the abundance of information that's relatively easy to access. We are right now working much more than people were working for most of the human history. Most people just don't have time to do research on various topics when they have to work and find time for their family. These AI tools makes the whole process much less time consuming. That's why the government want's to take it away from the people, so that they still would have a way to manipulate masses. Also they're all a bunch of filthy commies who at first said that AI will be the new means of production and than they seized it.
>>109135387Heh
>>109135387This is not needed for frontier models of china lmaoo
>>109160415>We are right now working much more than people were working for most of the human history.For most of human history there wasn't much to know, and most of what was known was wrong. A better comparison would be that up until 100 years ago, there was no real concept of a weekend. In 1926 Henry Ford changed the production schedules at his factories so that they didn't run on Saturday or Sunday, to encourage his workers to take up leisure activities. Across the whole US economy, the average number of hours worked per year per worker decreased from 2000 in 1950 to 1800 in 2010.
>>109160722And yet we are regularly quoting people who lived 2.5 thousand years ago. Its not that there was nothing to know, people didn't have the means to learn it until basically the modern times. And right now the government wants to control peoples access to the new kind of optimized education in the form of AI.
>>109162106People in general only want to educate themselves if they think it will make them rich or help them fit in with their in group better. Someone who watches CNN could easily change the channel or do a Google search for Fox News, to find out what stories they are not being told, and vice versa. The reason they aren't doing that isn't because we haven't "optimized" the switching process for them.
>>109148976Free markets don't exist.
>>109147993>silver is a rock with no inherent valueall goods and services have no inherent value, value is subjective>actually just as much as fiatsilver-backed currency is limited in supply, since the issuer of the currency ought to have enough silver to pay out to all his creditors should they redeem their bank notes. the market value of the silver is a function of the supply and demand for silver, neither of which is easily manipulated by the issuer of the currency. So silver-backed currency functions as a store of value, which is one of the functions of money.fiat currency has no backing, there is nothing to redeem. the value of fiat currency is only related to the supply and demand of the currency itself. supply of fiat currency is unlimited; there's nothing stopping the issuer from simply issuing more currency, thereby devaluing the debts he owes to his creditors. so fiat currency doesn't function as a store of value and therefore fails the criteria to be "real money".the functions of money are: a store of value, a medium of exchange, a unit of account, and a method of deferred payment. fiat currency arguably fails to fulfill the first and third functions because it can be devalued so easily
>>109163778People can and do increase the supply of silver all the time, and there's nothing in principle stopping a fiat currency issuer from maintaining a fixed supply of their own currency (or only increasing the amount of currency at a rate lower than the rate at which silver is usually mined). The reason to prefer silver is that the mechanisms and incentives for inflating its supply tend to be more robust against mismanagement than most fiat currencies, but if Elon found an asteroid with a quadrillion dollars worth of silver in it, the fact that this supply is still "limited" wouldn't be much comfort to the people holding silver.
half this thread are ai posts btw
>>109135387>Donald Trump says you're not allowed to use Claude Fable or GPT 5.6Good, let them go bankrupt.
>>109135387Okay. I'll just use one of the LLMs trained on claude data that will be released soon.
Everyone who disagrees with me is an AI. Because AIs are smarter than humans and only an idiot would agree with me.
>>109164058I wonder which of the orgs approved to access mythos are secretly full of chinese spies
>>109164162we can't even rule out anthropic and the nsa having them
>>109135387they paid Donald to do this to promote their company prior to IPO
>>109164409how does an unpredictable reduction in addressable market increase investor confidence? use your brain.
