On April 6, 2026, Sam Altman published thirteen pages titled "Industrial Policy for the Intelligence Age." The document proposes a national public wealth fund, a four-day work week, a corporate tax to protect social programs, and access to AI as a basic right. It sounds reasonable. Progressive, even. Bold, coming from the CEO of the company accelerating the very transformation the text seeks to mitigate.
That same day, The New Yorker revealed the results of a year-and-a-half investigation questioning Altman's credibility on AI safety issues. The timing matters. It defines the document more than its content.
What Altman presented isn't just public policy. It's narrative management, wrapped in flawless presentation. I've seen this pattern in other contexts: organizations defending their operations by disguising them as altruism. It's not conspiracy. It's structural logic.
Remember the case of Jensen Huang, who redefined AGI with a financial metric that favored NVIDIA. Now Altman designs the regulatory framework for OpenAI. The pattern repeats: whoever provides the tools declares the discovery and sets the rules of the game. Companies protect their operating conditions, and they do so by presenting it as common benefit.
Researchers at the Carnegie Endowment for International Peace summed it up bluntly: they call the document "communications work to provide cover for regulatory nihilism." OpenAI suggests structural changes that require decades of political coordination—a sovereign fund, tax reforms, labor redesign—while operating today without real limits. That asymmetry in time is the core of the approach. You promise a controlled future to excuse an unfettered present.
What's absent from the document reveals as much as what's present. Independent verification is missing. There are no mandatory deadlines or consequences for non-compliance. This dynamic appears in other contexts: the Chapultepec Principles on press freedom, corporate sustainability declarations. Rules without concrete human processes remain aspirations. Or shields.
The paradox is that Altman admits what his company is building: superintelligence that will cause massive unemployment and extreme wealth concentration. Such diagnostic honesty is rare in the sector. The flaw lies in the solution: a public fund and a shorter work week presuppose that institutions that ignored inequality for decades will now reverse it, on a larger scale. It's the same obstacle that appears in debates about universal basic income: redistribution depends on the willingness of those who benefit from the status quo.
The sector landscape adds layers. Anthropic, which emphasizes AI safety, recently suffered a leak of its source code. Its main rival responds with a responsibility manifesto. The entire ecosystem tends to its image while the technical foundation remains opaque. It's not personal hypocrisy. It's a rational response to imminent regulation they'd prefer to shape themselves.
This pattern has deep roots. Mesopotamian scribes monopolized cuneiform writing: they decided access to knowledge, interpretation of contracts, and which stories survived. When democratization threatened that monopoly, they proposed their own controls. The printing press in Europe saw guilds of copyists and the Church regulating what was printed and who authorized it. Technology evolves. Power over it replicates.
What would make an AI framework credible? Not just the content, but the authorship and design. It wouldn't be drafted by the entity most benefited. It would include audits by neutral third parties, deadlines with real sanctions, and radical transparency as a structural requirement, not as an advertising slogan. Because any promise of equity dissolves if the political winds change.
I don't reject all the document's ideas. Some open genuine debate. The four-day week has solid empirical evidence. Sovereign funds work in Norway and similar models. Access to AI as a basic right deserves exploration. The problem isn't the proposals. It's who states them, when, and what guarantees they won't serve—as Carnegie warns—as mere cover to operate without restraints while the debate drags on.
Researchers warn that AI's greatest risk isn't wild superintelligence, but superintelligence controlled by few: powerful systems in the hands of those who set their own rules and safety standards. Altman's plan, intentionally or not, falls into that category. It proposes future equity managed by those who prefer the present intact.
There are precedents of community governance that balance power. They're not perfect or universal, but they demonstrate it's possible to design structures where ideas prove their worth in practice, not just on paper. What's missing isn't imagination. It's the willingness to cede control.
The question persists: how do we ensure that good intentions translate into verifiable actions?
Stones don't lie, but historians sometimes do.
