Indeed, I find quite ironic that some people in tech in the US complain about EU "regulations first" approach, but then their government seem to arbitrarily stop things from being released because, well, there is no established policy on safety guarantees or other similar aspects.

For some conspirational reasons, I am thinking that this was the part of the deal between Anthropic and US government, all buzz was for PR, but behind the closed doors Anthropic asked US to regulate the LLM space

Well, when the leaders of this movement go around doom-trolling for years on end this is what happens. It turns out you need to be careful what you say if you're a highly visible public figure. Amazing!

Honestly, with open source models I don't think this regulation means anything because there's no way they can really regulate what's coming out of china. I don't think this affects innovation in AI much at all (unless your definition of innovation is "pour more money into diminishing scaling"). It's mostly just bad news for the US frontier labs, and based on their behavior I don't feel sad for them AT ALL. Like, they've basically alienated the vast majority of people by outright threatening their livelihoods or even society at large, and now we're supposed to feel sorry for them because they can't just go around saying "THIS WILL REPLACE ALL JOBS IN A MONTH!" without consequences?

I remember when Republicans told us they want less regulation and smaller federal government. Now they want their dementia riddled god king to control everything from pool liners to the information you're allowed to see, which is all in books and readily available online.

I don't see how you can make a case for a $700bn+ IPO when the government might not even let you sell your product. America is ceding the lead in the AI race. The winners here will be the Chinese AI companies.

If the Chinese models remain predominantly open source then it would probably be for the best. Unfortunately I'm not convinced they will, with examples like Qwen Max showing what could happen.

It will just mean US providers will rapidly loose their moat. Their moat is already shrinking. If they can't release their best models, it'll shrink a whole lot faster...

This is what OpenAI/Anthropic want, it's better marketing than they can pay for -- and it creates a precedent for permanently banning the next generation of open weights models

Not a fan of the phased release but I do remember when access to gpt-3 was gated and access to gpt-4 had a staged release.

ppl are acting like limited release is unprecedented when, in fact, has been the norm until a few years ago.

That only happens in governments that treat regulations as a racket, not something to be used for public good.

I wonder what kind of scheme the administration is up to. The obvious play is a squeeze where OpenAI and Anthropic are forced to give parts of their company away, like Intel. But they could also be toying with the idea of limiting frontier AI access to companies that bend the knee, which would further cement their grasp on the tech industry.

> There's been no public policy, executive order, legislation, or otherwise on this

I mean, insofar as you could frame each new model as its own patentable invention, "patent secrecy" would be an existing policy framework that clearly justifies what's been happening here.

1. Some private company or individual invents something.

2. But the state wants a monopoly on the new thing's military use-cases (against other states/militaries.) So the state forcibly classifies the invention at patent time.

3. But the individual/company still wants to make more money than the state is giving them as compensation for their own use of the patent. They want to sell to the private market. But, of course, the state doesn't want to permit this, as arbitrary private parties could in turn resell to foreign state actors.

4. So a compromise is struck: private use is now made deny-by-default. The state permits the individual or company to sell into the private market, given: a rigorous per-customer KYB background-checking process; strong supply-chain tracing; contractual stipulations prohibiting resale; and the customer use-case being transformative or cost-prohibitive to extract the original invention from.

5. As such, big established enterprise customers who want the invention for private use in their internal industrial processes, can somewhat-easily jump all those hoops to acquire access to it; but everyone else is now basically locked out.

Sound familiar?

So OAI are you also silently dumbing down your models when you detect "inappropriate topics" like Anthropic did with Fable?

I largely blame people like Amodei for such outcome. As product owners, they could've done it the old way: telling people how great this product is, how much potential it has, and what kind of guardrails the companies are building and etc. But oh no, Amodei has to do the doom trolling 24x7, while in the meantime plays a cult leader by telling people only he knows how to the guarding angel of the AI or the humanity thereof. Ironically, the same people also push their companies to develop more powerful AI in full speed. They think ordinary Americans are so stupid that they can't see through them?

The precedent for this is terrible.

MAGA is bad enough. Imagine if the current batch of US progressives, who have 0 idea how any of this works, wins the presidency and gets to decide who gets to use it.

I agree that this is all ridiculously arbitrary right now, but it shouldn't be surprising either.

I can't find the exact blog post (maybe on simonwillison.net ?) but I read people predicting that know your customer laws would be coming to AI if it gets more powerful several years ago already.

Powerful technology that can do immense harm in the hand of individuals/small groups is the most obvious (and legitimate!) target for regulation. Maybe Anthropics hubbub around Mythos made all of this happen earlier than it would have, but it was going to happen (if the models are going to get as capable as valuations imply they will).

(Edit: Of course this doesn't mean it can be effectively regulated in practice)

Further more, no one actually gets hurt if we start rolling these things out more slowly.

Rolling them out selectively according to the whims of an administration that disdains fair process, tears down the institutions that could potentially provide and legitimise reasonable regulation, etc... well, that's another topic.

In the EU that's the norm not the exception. A little taste of Europe for our American friends :-S

How can you simultaneously be a bottleneck for innovation while being their largest customer, and pouring tons of money and resources into it to help accelerate development?

The startup-brained among us never learned the first rule of business which is to not fuck over your benefactor.

I'm going to get downvoted here, but all the E/acc people that loudly allocated for Trump, someone known for amassing power by any means necessary including strong arming industry should be publicly eating crow right now. This was something that was always in the cards when you vote for someone who only cares about himself.

The big companies want this. It's a moat for them, a way to keep competitors (especially overseas) out of the US market.

They might try to extend it to downloadable open weight models, but honestly they might not even bother with that. The goal is to keep people from competing on lucrative contracts or the hosting market.

Shadow of export controls is very long indeed.

The Project is almost here.

It's worrying that with no formal and transparent policy framework that the government will be picking winners and losers and stifling innovation.

The market will demand such a framework. I suspect that's the larger idea here, in that Amodei not only wants to be in the room when that framework is written, he wants to be at the head of the table.

He apparently wants it so badly he's willing to set back his own company's IPO to make it happen, given that there can be no pure-play AI IPOs until the regulatory picture is sorted out.

But think about how terrible it would be if “foreigners” (including the ones that work on these models) got access!

We must clutch our pearls and cite National Security as a reason to pick winners and losers, just like the government did for Fable.

I'm honestly surprised there isn't more political outcry. The administration has a party affiliation that, typically, insists on free market principles and is against government overreach and regulation.

You would think that this government, attempting to puppeteer the most rapidly growing industry in the world, would have more people outraged.

Where are all of the people crying "Communist"? This is one of those moments where it is less of an overreaction.

[dead]

Just because you don't get access, doesn't mean they're not innovating.

This move was obvious the moment Anthropic pleaded to the government to regulate them.

As predicted, [0] it has now been applied to OpenAI and soon anyone else releasing highly capable models.

[0] https://news.ycombinator.com/item?id=48511849

[flagged]