As far as Europe is concerned they have recently signed up to the 'Pax Silica'[1] and willingly givrn the LLM space over to the US incumbents with buildtin legislation banning Chinese models and coperation with them. So EU will be a renter of the LLMs that the US allows them to use. In the long run OpenSource will dominate as it did in the DB(MySQL/Postgres)/ServerOS(Linux/BSDs) versus Proprietery rent seeking alts like Oracle and Microsoft et al. Would be interesting to see what the global startups using Qwen/DS/Kimi etc within the EU-US space navigate the cutting edge OpenSource LLMs vs seeking/getting a permission slip from the US gov.

[1]https://archive.is/aiJiq

> I can train insainly powerful models on my laptop.

What's with this hallucination? The thread is about GPT-5.6. Your laptop can't even run gemma 4 unquantized bfloat16, which is light years behind GPT-5.6, and running it is light years from training it. If something that a laptop can train is insanely powerful for you, you don't need to worry about this thread at all.

This also completely screws over U.S. businesses. American startups will be forced to pay premium prices for nerfed, heavily-censored, 'compliant' models from a few massive corporations. Meanwhile, foreign startups will be running cheaper, unrestricted open-source AI. We'll price ourselves right out of the global market.

What government wants to have their population use foreign AI. (Not many). Only issue I see is good enough is what the majority might be okay with.

So far it's only US doing this. I don't think it's in anyone else's interest to limit development of open source models or chips. Nvidia has secured a leading position in GPU market by being the best overall, but if US continues to mess up with the export, that changes the calculation and surely we'll see the alternatives

> This will make it hard/impossible for new vendors to come into the market and only established companies will get to play, and charge, for LLMs.

I don't understand how you leap from "US govt. decides who gets to use GPT-5" to "limit new vendors from competing with OpenAI".

Can you walk me through the logic?

Of course. The US government inserts itself into commerce, starts picking winners and losers, and thus creates an industry for petitioning gatekeepers at all levels for access. The goodfellas profit and the public pays. OpenAI and company get manufactured scarcity to raise margins, and lawmakers get to collect an entry fee. This was doomed to happen. The US is much closer to a total state than most give it credit for.

>This is regulatory capture in action.

Isnt this all export control based? If so its not regulatory capture for a few reasons. If not disregard this.

1) new entrants wont get export controlled because they arent leading edge

2) a new company could just implement KYC. It could even be a competitive advantage (Anthropic wont or cant)

I guess if the new model has the capability to do 'something' that's national security threat, then this makes sense. Otherwise this move makes zero sense, and actually is a drag on innovation - who wants to invest money and people to make a better model that can't actually be sufficiently sold to make a return on investment? Meanwhile other models from Europe or China that are better steal market share. Though that's not to say they won't do the same thing for the same reasons.

> Are we heading to a world where GPU use is regulated

Well, there was the kerfuffle around PS3 (IIRC) and 'supercomputers'. I suppose that would be the 2020s version of that. What's old is new again. Ultimately you can just continually wire together less powerful hardware to come up with something that can do the job.

As I understood they will limit running those undesired llms with Remote Attestation and Trusted Execution Environments.

>can train insainly powerful models on my laptop

Can you give an example or two?

This isn’t about keeping people from having the power of frontier LLMs. So tricks that let others have it aren’t a defeat of this policy.

This is export control, where the US government seeks to leverage the fact that these frontier models are US made. This is then leveraged against opponents, and likely also just for grift. There’s also perhaps a little legitimate worry about the implications of free access to, but that is secondary to the real goal.

> I can train insainly powerful models on my laptop.

This is such hyperbole. You might be able to train a model that's merely useful in a single domain, but to say you can train an 'insanely powerful' model on consumer hardware is laughable.

Warm up your VPN to zAI for the eventual banned GLM-6 I guess

What? This is the opposite of regulatory capture. Neither anthropic nor openai are getting to choose what happens with their models.

How is this regulatory capture? Any new LLM company can just exist outside of the US and export to everyone.

literally the opposite of regulatory capture - it spells trouble for specifically openai and anthropic mostly

> I can train insainly powerful models on my laptop

Can you?

>I can train insainly powerful models on my laptop

Explain. The labs have been spending about $100 million in compute to train a model.

Nick Bostrom wondered aloud in Superintelligence (2014) why states would allow individuals and private organizations to develop AGI. If one takes the possibility seriously, AGI would a source of immense power and any state would to take that opportunity for itself.

Edit: Not saying whether AGI is right around the corner, that's a different discussion. I'm just saying that a serious possibility of AGI and an understanding of possible consequences will make a state act.

> This is regulatory capture in action.

With the twist that it will end up involving payments that directly benefit Trump, following the Mafia business model that he learned in the construction industry and that he's brought to the White House.

GPUs and computer hardware prices have been on the rise. I can see a twisted perspective where it’s stated that the US government needs to closely control computer hardware that can run particular LLMs, as a national security interest. At least that idea isn’t completely wild now seeing what we have been experiencing.

Remember those weird conspiracies we used to have about universal surveillance; tracking and so forth? Well if you think back to those and whatever might happen with GPUs and hardware, and LLM restrictions or the likely age gating//ID’ing that is to come from this, that’s a good guiding framework for how this will proceed and affect normal people.

Buy Bitcoin.

look, i'm sorry, but this is a thoroughly solved problem by now.

> Are we heading to a world where GPU use is regulated to ensure that illegal LLMs aren't being processed on your machine?

Actually that sounds pretty reasonable considering current regulations regarding almost any other important resource / material that affect the general population.