Open Weights and American AI Leadership
10 points by stick
10 points by stick
I'm finding the list of signatories on this letter really interesting and somewhat confusing, looking purely through the lens of economic interest.
It makes sense to see Meta, the startups, and the VC firms on this list. And it makes sense to see OpenAI, Anthropic, and Google all missing. Microsoft itself is somewhat unexpected to me. I don't think of them as having a focus on open weight models (no more than Google), and they have a large stake in OpenAI. Maybe they're looking at this from the angle of Azure providing compute. But then why is Amazon missing, when it has AWS?
And of course, we don't see DeepSeek, Moonshot, or Z.ai on here. The letter is about American technological leadership after all. But then we do see the French company Mistral! Mistral who released a playbook for Europe becoming a self-reliant AI powerhouse.
I assume all of this is in the context of influencing the Trump administration's thinking on Chinese open weights models. But the letter is really about American open weights models. The call to action is all about keeping the American open weights ecosystem competitive. I didn't even realize that was under threat, so I feel like I'm missing something.
Microsoft itself is somewhat unexpected to me. I don't think of them as having a focus on open weight models (no more than Google), and they have a large stake in OpenAI.
To me it very much aligns with the trajectory of MIcrosoft cooling on "winning the model race" through their investment in OpenAI and towards a mode of "we don't care who wins, we'll make money off cloud".
But then why is Amazon missing, when it has AWS?
This is actually the more confusing part IMO: AWS is also in the business of "we don't care which model wins" and from what I hear, they don't seem to be tying themselves to a particular model for e.g. their engineers and internal tools.
But then we do see the French company Mistral! Mistral who released a playbook for Europe becoming a self-reliant AI powerhouse.
The way I see it, the US is the still ~biggest customer for Mistral as it exists today and it would be a big blow to their revenue if e.g. open weight models could not longer be used by medium-large companies in the US i.e. via a ban on companies contracting with the US government from using the models.
This part of the letter:
In shaping this ecosystem, policymakers should be careful not to conflate legitimate model-development techniques with misappropriation. Distillation, or the practice of using one model’s outputs to help train or improve another, is a widely used technique for model improvement, evaluation, and validation. It reflects a long tradition of learning from, building upon, and improving existing technologies, a tradition that has helped drive innovation since the rise of the open-source software movement. By contrast, unlawful efforts to extract value from closed models raise legitimate concerns. Those concerns should be addressed through targeted legal and commercial frameworks rather than sweeping restrictions on techniques that play an important role in AI innovation.
Made me think of the old halloween documents they accidentally published in the 1990s.