Whose Future Is It, When One Country Writes the Rules?
Sam Altman's proposed US-led AI safety forum raises a sharper question than safety: who gets to author the rules for a technology meant for everyone?
When Sam Altman proposes a global forum for AI safety that the United States would lead, the real question isn't safety. It's authorship. A technology sold as belonging to everyone would have its rules drafted by one country, and the gap between those two facts is where the next decade of power gets decided.
Picture the moment as Altman frames it in his Financial Times op-ed: an international body modeled on the IAEA, the agency built after the atomic age to keep nuclear power from becoming nuclear catastrophe. Nations gather. Standards get set. Safety becomes a shared language. It sounds like cooperation. Read it twice, though, and a smaller word does the heavy lifting: led. The forum would be led by the US. Everyone is invited. Someone owns the room.
That distinction matters more than it looks. There is a difference between a table where everyone sits as equals and a table where one nation sets the agenda and the rest respond. The IAEA analogy is telling in a way Altman may not have intended, because the nuclear order it governs has never been equal. A handful of states hold the weapons and write the rules about who else may. The proposal, described in the original report from SiliconANGLE, imports that structure and calls it fairness.
The United Nations has been making a quieter counter-plea for months now, arguing that a technology reshaping every economy on earth should be governed by a body where the smallest country has a voice, not just the largest. It is a plea from the passenger seat. And the plea reveals the stakes plainly. Whoever chairs the meetings decides which harms count, which uses get blessed, whose values are baked into the defaults the rest of the world inherits.
Consider who inherits those defaults. A crop advisor built on a foreign model tells a farmer in Kenya when to plant, trained on weather patterns and market logic from somewhere else entirely. A teacher in Manila runs her lesson plans through a system whose sense of a good answer was tuned in California. A nurse in São Paulo trusts a diagnostic tool whose thresholds for risk were set by people she will never meet. None of them attended the forum. All of them live inside its decisions.
This is the part the debate on X tends to miss when it splits into cheering and jeering over "America First" AI governance. The critics call the US-led framing a power grab, and they're not wrong about the direction. But the defenders have a point that's harder to dismiss. AI is power, economic and military and cultural at once, and no nation ahead in a race hands that lead to a committee of equals out of principle. That isn't a conspiracy. It's gravity. The country that builds the frontier will want to keep building it.
So the tension is real and it doesn't resolve cleanly. You can want a genuine global standard for AI safety and also understand why the country holding the lead has no incentive to share the pen. You can want both. The Altman proposal reportedly floated a 5% stake concept — a figure that circulated without a clear owner attached, which is exactly why it should be treated as a trial balloon rather than a plan. What gives the balloon weight is the instinct behind it: safety offered to everyone, control retained by one.
The word governance makes all of this sound procedural, a matter of committees and communiqués. It isn't procedural. It's the slow decision about whose assumptions become the water the rest of us swim in. The farmer, the teacher, the nurse won't read the charter. They'll just wake up one morning inside a set of rules they had no hand in writing, wondering why the machine that runs their day seems to understand a world that isn't quite theirs.
FAQ
Why does Altman want a US-led AI forum instead of a UN one?
Altman's Financial Times op-ed frames a US-led body modeled on the IAEA as the fastest route to global AI safety standards while keeping the United States at the front of the race. A UN body would spread authorship across all member states; a US-led forum keeps the drafting pen with the country already ahead.
What is the IAEA comparison actually saying?
The IAEA was created to govern nuclear technology after WWII, but its order has never been equal — a few states hold the weapons and set the terms for everyone else. Applying that model to AI imports the same asymmetry, which is why critics read "cooperation" as consolidation.
Who is affected by AI governance decisions they don't participate in?
Everyone who uses AI tools built elsewhere: a farmer relying on a foreign crop advisor, a teacher running lessons through an overseas model, a nurse trusting a diagnostic whose risk thresholds were set abroad. They live inside governance decisions made in rooms they never entered.