Last week the U.S. government did something that I think will reshape the next decade of AI. After a jailbreak exposed the cyber capabilities underneath Anthropic's newest models, Washington used export controls to bar Fable 5 and Mythos 5 from any foreign national. Not just people abroad, but non-citizens inside the United States, including Anthropic's own employees. The company had little choice but to pull the models for everyone. In a single directive, access to the most powerful AI on earth was sorted by passport.

That is the future arriving early. We have treated frontier models as global products, available to anyone with a credit card and a connection - last week’s events ended that assumption. Once a government decides a model is a strategic asset, the next step is obvious. Nations will not want to depend on another country's lab, export rules, and idea of who is allowed in. They will want their own. Sovereign AI is no longer a conference talking point. It is a national security requirement, and the race is underway.

There is a second force accelerating all of this: Open Source.

The same week as the Fable 5 ban, Z.ai released GLM-5.2 with open weights under an MIT license, and it beat GPT-5.5 on real-world coding benchmarks at a fraction of the cost. The gap between the best closed and open models has narrowed to the point where a capable nation no longer needs a hundred billion dollars and a frontier lab. 

A sovereign model is only half the equation. A country that controls its AI will want to decide who uses it and in what order: citizens first, domestic businesses next, maybe strategic partners after that. That only works if you can actually prove who someone is. This is why proof of human and tamper-proof proof of nationality move from nice-to-have to foundational. The building block already exists. Most governments issue a cryptographically signed passport, and that signature can be verified with zero-knowledge proofs that confirm someone is a real, unique human of a given nationality without exposing their underlying data. This is exactly what we work on at Self Labs. 

I want to be honest about why this matters to me. I want a world where everyone everywhere can reach the best AI, where a builder in Nairobi has the same economic opportunity as one in San Francisco. Self powers bold ideas such as universal basic compute precisely because that access should not depend on where you were born. But in a world where export controls become a norm, the next best thing we can do is make sure that when people prove who they are to reach these models, their privacy is preserved. Far better to access a model through a zero-knowledge proof than to hand over your passport to a KYC vendor and have your data stored, sold, or leaked.