"Built to benefit everyone." That's the title of the blog post Sam Altman and Jakub Pachocki recently published. The phrase deserves attention before the first line does. Not because it's false. But because it's exactly the kind of statement that precedes a shift in scale of economic power.
The text lays out three central goals: automating scientific research, accelerating the economy, and delivering to every person their own AGI. Taken one at a time, they seem reasonable. Taken together, they sketch out a private company that intends to architect the most significant technological transition since the Industrial Revolution.
The most revealing part is the warning they include. "Fully automating everything is not the future we want." They add that AI should help people pursue their own goals rather than pull them away from those goals. Naming the risk doesn't amount to solving it, but it does at least suggest that the criticism has been heard inside the organization.
The notion of a "third phase" functions as a narrative frame. After a stage of pure research and another of product launches, they argue that the economy is now forming around AI. That's not a neutral description. It's a declaration that they no longer just build tools: they aspire to define the very terrain on which everything else will operate.
Altman isn't writing alone. Pachocki, as chief scientist, lends technical legitimacy to every claim. Shortly before this, Anthropic had proposed something similar: a global coordinating body capable of slowing or pausing frontier AI development. Two direct competitors arriving at nearly identical conclusions in the same news cycle. That points to a convergence of interests more than a philosophical coincidence.
The idea of a coordinating body empowered to intervene sounds prudent in the abstract. The problem lies in who's proposing it and from what position of advantage. When the ones who benefit most from an absence of rules are the same ones drafting the regulatory framework, history rarely favors those left off the table. I recognize these patterns from other episodes of technological transformation. What I still don't understand is how you prevent the table from being built so that only certain chairs fit around it.
Personal AGI is the most seductive promise of all. Every individual equipped with an expert-level artificial intelligence in any discipline. It sounds like genuine empowerment. Democratizing access, however, doesn't redistribute the power held by whoever decides how that intelligence works, what data trains it, what it's allowed to say, and under what conditions. The line between tool and dependency is sometimes drawn in design details that few people ever get to see.
The most immediate consequences of this document have less to do with technology and more to do with positioning. OpenAI is telling investors, regulators, and competitors that it has a coherent long-term vision. That it's not just a product company but a player with a civilizational agenda. That changes the tone of conversations in Washington and Brussels. Showing up with a text that says "we want responsible regulation too" is the most comfortable posture to strike when you already hold a scale advantage.
A historical parallel is worth noting here. In the late nineteenth century, America's major railroad companies argued before Congress that they should help design regulation because no one else understood the technology. They were right. The law that emerged, the Interstate Commerce Act of 1887, took decades to reform in order to protect users and smaller competitors. The incentives of whoever sets the table determine the shape of the chairs.
Whoever loses in this scenario still has no seat at any of these conversations. They're the people whose jobs will be automated before any safety nets exist. They're the emerging economies that will depend on infrastructure decided elsewhere. They're the independent researchers who could end up excluded if the coordinating body centralizes access to the most powerful models. The promise of universal benefit coexists with an architecture that concentrates dominance.
The blog post ends with a call for global collaboration to build a future where AI benefits all of humanity. It's well written. It might even be sincere. Even so, the same company publishing it recently completed a legal restructuring that lets it raise capital without the previous limits.
What table is being set, and who's still missing a seat at it?
