alt Hacker NewsU.S. government will decide who gets to use GPT-5.6
Comments
LLMs have two avenues. One is the realm of self improvement in fields with verifiable output like programming and math. The other is natural language where they can't generalize super well, and therefore need a lot of new training data. Today new data means interactions with an LLM, which is what only the leading chat providers have. In both cases they will continue to improve and slowly replace white collar jobs. The biggest bottleneck currently is lack of hearing and seeing capabilities which isolates LLM training data input to entered text mostly. Once they start interacting with people by observing their behavior, almost all knowledge will be trained on. Then they will become adept lawyers and psychologists. Behavioural understanding can be only 5 years away. After that they will be limited by their navigation , i.e., robotics.
Cyberpunk 2077 is such an accurate picture of the future. Megacorps will own and control everything, we will be lucky to get the leftover scraps.
The downside of marketing your product as being too powerful to be safe is that people start taking it seriously. I’m surprised there isn’t more discussion about the contradiction. If it really is the end of labor and a deadly weapon then export controls actually make sense.
So which position should the government take? The one it’s doing? Or that LLMs aren’t actually key to national security.
Dictatorship. Complete with a trashy despot. Coming to a country near you.
I'm finding it extremely hard not to have a cynical perspective on all of this. There's an idea that I've been mapping onto this whole this, which could be called something like effective knowledge. Regular old knowledge is just information, or access to information. Effective knowledge is the integration of all that information into an understanding that can be acted upon. That requires things like time, money, and involves the usual socioeconomic hurdles that have separated people into groups like "laborers" and "knowledge workers". Sure, in theory "anyone" could read textbooks and learn, but only a select few have the time, money, mentors in their lives, and so forth to really do that.
The rise in capability of LLMs over the past year has basically removed a lot of these boundaries for people. Learning, building, and experimenting is a lot easier when you have a capable partner like Claude to help you along the way. Claude doesn't always get everything right, and you have to be a skeptic, but it's a lot better than nothing.
When I see the government restricting access to LLMs (or Anthropic as they were doing with Mythos before the whole Fable debacle), I basically just see the same old pattern of the ruling class moving to protect their advantage by keeping the great masses in ignorance. Broadening access to LLMs (i.e. effective knowledge) would put everyone on a more level playing field. But we can't have that, because politics, nations, the economy, blah blah reasons reasons. Guess utopia will just have to wait a bit longer.
But then again, this feels a lot like cryptography export controls. Those controls are in place, but I doubt anyone really thinks they work or make much of a difference. Software is not like nuclear weapons, and a data center is a much smaller lift than a Uranium enrichment facility. So maybe this is just a temporary roadblock. But let me tell you, I sure am ready for it to feel more like the government is working for (not against) the people.
The new normal for the next decade: You must protect the public from us and all others, and we are your closest ally so we make your rules for us and our competition. This wasn’t a lucky outcome. They laid the groundwork years ago with the AI “ethics” movement and this was the play all along.
Whether or not you assume bad intent of the government here, the amount of money the export control on Fable has cost Anthropic has to be unbelievably high
Eg I expect I would have paid more than 2x per day what I spent the past few weeks, and if gpt 5.6 comes out and is competitive that's going to absolutely gain market share.
An unbelievably costly turn of events for them
To all of those thinking that GLM/OSS will save you — keep in mind that the model size needed to compete here likely requires an NVL72 or similar — 72GPU dedicated infra to run a hosted model. This will almost certainly get regulated by the gov’t as well, and even if not so there will only be a handful of companies that can afford it.
Remember when you could only get the Netscape version with good encryption if you were in the US, because the government had classified encryption as a munition? And that the rest of the world had no trouble matching or exceeding that level of encryption?
Is this going to be like that?
Do we know how much choice OpenAI has with the arrangement? They call it a "request", but could they have been ordered directly?
Will these ad hoc decisions by the U.S. government, without law or clear process, not hurt the coming IPO's of Anthropic and OpenAI?
I guess Orpheus fined 50% of Eurydice's soul for early withdrawal attempt - No take backsies is clearly in the Contract.Upheld.But Orpheus gets a coupon for free ressurection (Void in thrace:There can be sage).
The irony to have the EU criticised for regulation on one side but complete government control of access on the other.
Damn. This is the second post today that just disappeared from the top of the landing page. This is 100% manipulated.
So as a European, I'm being blocked from all new models apparently. I'm a big fan of using Claude Code for my sideprojects and for those I don't really care about sharing context. Is there aything that comes close to Claude Code and is still affordable?
> while AI companies and the administration work out a longer-term plan for regulation on the sector.
It’s not really the executives job or role to create new regulatory structures. If they want something durable, that lasts more than one administration, they need actual laws passed by Congress.
I think there can be smarter ways to fix security issues right. Like you let AI loose in a gated environment fix all things it flags and than release the model ? Any new changes you make you now anyway have the AI vet it ?
This post isn't even on the landing page for some reason.
I guess we need to kick start SETI@Home like compute donations for training models.
"All publicity is good publicity"
Whilst this policy is driving countries around the world to develop new AI strategies as USA cannot be relied upon, on the plus side for OpenAI, the publicity of this will help drive customer sales.
Well, as long as the government is deciding, that’s alright then. The US government is a paragon of incorruptible integrity and even-handed, thoroughly considered reasoning. We’re in good hands, folks.
It's pretty easy to solve. You just keep pushing new versions of Opus 4.8....
Fair enough, i'm off to use GLM. Let's see how this plays out US gov
The rest of the word at some point will increase chinese AI company. I am using some of their services and for many task they are good enough . It can be a winning strategy for a short period and a disaster in long term. Xi and China i think , are very happy of these decisions. They are building their model and their hardware. Even if for now it is subpar ( i don't know, it is an hypotesys), money that us will lose for this choice will make Chinese products improve a lot .
There is an assumption that everyone is making here - that China will not do the same. It is entirely possible, that China restricts their frontier models - as and when they are developed - to only Chinese citizens. And India follows along.
IMO AI is different from everything else. It is a weapon as potent as nuclear. It is only natural that it be treated as one.
> We don’t believe this kind of government access process should become the long-term default.
Really glad to see some reasonably prominent pushback against this government overreach.
The information has been reporting that the government wants to individually approve which companies get access and when.
Imagine the wonderful opportunities for corruption and influence peddling, not to mention, excluding any companies that don’t support Trump
Non-paywall: https://archive.is/PCQQl
Great so we just need to wait for China to catch up I guess
I can't tell if this is bad for the big labs, or good because it means they now have an excuse for not showing meaningful progress in the lead ups to their IPOs.
Does anyone else think this is all just FUD, smoke, mirrors and marketing hype? "Look, our model is so good that the government told us to stop" "Our model is so good that the government is going to control who has access to it".
Come on.
I think occam's razor can be applied here. And like everything else, its about money. I don't know exactly what the play will be here, but this doesn't sound like this technology is too powerful and more like billions of $ lost investments need to be made up somehow without the people getting annoyed about the government bailing these guys (AI companies, investors, etc.) out.
David Sacks has been silent for a long time.. So much for being the big “AI czar”. Does he have any influence left in the government?
This makes no sense, it only will embolden any attempts by china and other countries to move away from depending on US AI tech
Without access to leading models, I think open source LLM development will also slow down. I'm not sure which portion of their success right now is due to RLAF and distillation but it's certainly not zero.
Superintelligence for me and not for thee.
What are the odds this is going to become another avenue for grift - magically any companies the trump family invests in are going to get access. Any companies that aren't sufficiently 'loyal' to the regime will have to wait or may never get access.
Even if these models are so strong that they become a security threat if in the wrong hands this arbitrary and intransparent regulation worries me. Not for the current SOTA model vendors but for the open weight models.
I have always wondered how AI industry is not in a (partial) bubble because open weight models will substantially eat into their revenue (rather sooner than later). But with this ('bro') regulation it seems that an easy strategy could be to regulate this problem away, especially easy to convince everyone that those chines models are truly evil.
Wowzers. It's been some time since export controls were something i'd see in software. Interesting times.
Serious question, what are my alternatives?
All they had to do is to keep their mouths shut.
Here's an unpopular opinion: this might be the only way to deploy advanced models. A lot of people compare advanced AI with nuclear weapons. Creating white lists of users that are allowed to use advanced AI feels wrong. It feels against everything that the Constitution stands for. That men are created equal and they are free to pursue their happiness. Now they are free to pursue that happiness only if the US Government signs off on that. It hurts to only think of that. But I'm afraid there's no other way. These models, in the wrong hands, can result in unfathomable devastation.
How do I know? Dario Amodei said that when he explained why Anthropic has to limit the US Government's usage of its models [1]:
> Some uses are also simply outside the bounds of what today’s technology can safely and reliably do. Two such use cases [...]: Mass domestic surveillance. Fully autonomous weapons.
[1] https://www.anthropic.com/news/statement-department-of-war
In a scenario where some breakthrough in fusion energy will be discovered I envision:
- instant, total world war if it's not coming from USA
- let's finish all oil's reserve first otherwise
Less money to US AI tech. This could be good long term to move us away from them.
Can anyone explain to me (a non US citizen) how this won't be found to be unconstitutional (eventually)?. I would think it falls under freedom of expression. And given the attempted classification of encryption as a munition that failed, I don't see how this can possibly last?
Thanks to the US government for helping kill Anthropic and OpenAI by preventing them from recouping any R&D money from new models. Doing god's work.
OpenAI/Anthropic are begging to be restricted because it's great marketing and it creates a precedent to permanently ban open weights models. The problem is nobody in government believes in/cares about commodity pricing of truly open AI and how much it could help the world economy and prosperity.
Companies like Microsoft have been asleep at the wheel in terms of security for decades and now there's a model that can identify where they've been careless. That's not a "nuclear bomb level threat" or whatever Anthropic wants to call it, it's reckless carelessness by the existing companies.