|
If you think I've insinuated they don't have a motivation, you've misread. It's undeniable they have motivations (though I pointed out, the signatories and non-signatories all have the same motivations). From a PR perspective, the motivation would push toward being a signatory. My overall point is, don't let their motivations, nor your cynical conclusions based on those obvious motivations, blind you to your own point of view. The debate here shouldn't be about the signatories motivations, it should be about the risks of open-weights coming too close to the untested frontier. You've acknowledged that now: "I know that a powerful open weight model in malicious hands is worse than OpenAI doing a whoopsie". But I'm left a little confused about your position because you follow that up with what I can only describe as a set of "what-about"-isms. Yes, closed models could still be misused, especially in the wrong hands. Still, even worst case there, it's fewer hands. China is the only set of hands that's really relevant for now, no one else is close. Without going into deep detail, fairly sure we both don't trust the Chinese government. But we also don't have a lot of practical choice on whether they have access to a model frontier-ish. What could be different is if Russian hackers, Malaysian hackers, Iranian hackers, etc. all have access too. In the open-weight model, they sort of do (somewhat constrained by compute.. but rounding up enough compute for a bit of inference is achievable even for them). I should make it clear at this point that I think banning US companies from using open-weight models would be a totally ineffective policy that's just self-harm. Anthropic said the same and I agreed, but just in case that stance has become blurred, I'll re-clarify. It's probably more unclear because there really isn't a policy stance on the table that backs what the "maybe there shouldn't be frontier open-weight models", because any real and effective policy there would either have to be multi-lateral, or be the result of independent actions that just happens to align. So really, all that an independent actor like Anthropic can do is say "maybe there shouldn't be frontier open-weight models", which they did and see if that leads the world toward agreeing with that position, which might then lead to some action. If that somewhat tenuous outcome ever did occur, it wouldn't really mean a "moat", as the Moonshot, Zhipu, Alibaba, Bytedance would still have a model and offer it via API. And they could still create agreements with US/EU/etc. cloud companies to host them, in the same way as other closed-models are hosted. But there'd be more control. Yeah, there'd be some advantages to model providers, compute providers from this. There'd be some costs to potential customers and those that would self-deploy for good purposes. As little as I support those costs, and as much as I'd prefer not to hand out those advantages, my cynicism would not prevent me from realizing that it's a bigger problem that if you don't do this, you'll fail to harm the interests of those who would self-deploy for nefarious purposes. |
Very fair point, taken. So, setting aside Anthropic & Co's motivations: trying to restrict this is pointless without planetwide cooperation - I don't see China complying - and likely to just result in dividing people into haves and have-nots. Furthermore, allowing this access, nay, power, into the hands of the very crappy stewards around at the moment doesn't really protect anyone, just makes them richer. I'd stomach their endless lust for more money if it was just a side effect of making the world secure, but I don't see that happening either. What do you realistically do when China keeps advancing the SOTA? And really, all the other nations I mentioned - they're one leak (and a few H200s) away from acquiring the ability anyway. They don't need the ability to serve 250 million concurrent users to be dangerous, just a handful of instances in the hands of their government officers. Who are we withholding the knowledge from, at that point? For what benefit?
I won't even touch the point that this is.... ultimately, knowledge acquired by scraping the internet, so it should belong to everyone - and now they're even pissed off someone is "stealing it back" by distillation - nice discussion for another time.
Furthermore - this approach worked for nuclear weapons for a while as the material is really hard to source, supercomputers weren't available, centrifuges are hard, and good nuclear scientists were so few that they could be individually monitored or assassinated if they helped the wrong party (really). Fast forward to present day when there's no practical way to prevent NK or Iran from advancing their knowledge. We're trying the same obsolete approach, but it applies even less thanks to the Internet. Also, even then, a couple leaks in the 50s gave the soviet union parity - any parallels to our scenario?
I really appreciate your points, I agree with the danger, but I don't see any approach really fixing the situation, and those guys' proposal is just a get even richer quick scheme.
Finally, of all people, finally a day came where I find myself in full agreement with Zuck (https://www.nytimes.com/2026/07/28/technology/mark-zuckerber...). Obviously his motivation is also not lost on me - this is bad only because it's not him this time, but still, even a (morally) broken Zuck is occasionally right.