A language model that decides which version of the facts to show you. A government that makes money every time someone uses it. Camera glasses that millions will buy of their own free will. Each piece, seen on its own, looks like a logical step in technological development. Together, they point to something deeper.

OpenAI is preparing its next model, one the company claims rivals the market leaders. Instead of simply releasing it, the company submitted it for review by the U.S. government. That much was already known. What's new is that the company is proposing to hand over to the government a percentage of the revenue that model generates.

That's the twist. The government stops being an external regulator. It becomes a stakeholder in the commercial success of the very tool it's supposed to oversee.

The incentives change completely. When you receive direct benefits from something, it becomes harder to scrutinize it rigorously. What emerges is a tendency to protect and polish the product rather than subject it to real scrutiny. And when millions of people consult that model to understand historical events or current debates, its answers end up shaping perceptions. Who decides the tone on geopolitical issues or controversial policies? With a financial stake involved, the versions that drive the most adoption stand to win, whether or not they're the most accurate.

I've seen similar dynamics in how power gets exercised around emerging technologies. This move by OpenAI fits that pattern. The interest isn't only in limiting AI. It's in becoming part of it.

This reading is a personal one, but the pattern with Sam Altman is telling: solutions get promoted for problems that these same technologies deepen, like mass layoffs happening alongside enormous investments in AI. These are rational moves when conditions allow for them. After years of watching this kind of social experiment play out, the formula repeats with a consistency that no longer surprises.

Things get more interesting with Elon Musk and his Grok. If OpenAI is offering profit-sharing in exchange for regulatory smoothness, other players may look for similar arrangements. Musk maintains ties with the U.S. government and has already adjusted platforms according to his own views. A model built under that kind of scheme would reflect those same balances of power.

Models developed in France, like Mistral, or in China, like DeepSeek, operate under different incentives. They aren't neutral — all of them carry the imprint of their training data. But by not depending on Washington's approval, they retain room for alternative approaches. It's ironic that the rivalry between blocs, which almost always divides, ends up preserving less-aligned options here.

While language models follow this path, Meta is pushing its smart glasses. The fight against mass surveillance has changed shape. It's no longer imposed by mandate. It's adopted out of everyday convenience.

Every person who wears them becomes a capture point sending images to Meta's cloud. The people around them give no explicit permission whatsoever. The usefulness of the built-in AI is enough for people to line up, entirely of their own accord, to buy them.

This visual data infrastructure, combined with models fine-tuned through agreements with the government, brings to mind structures of informational dominance from other eras. The Romans controlled the grain supply and trade routes. The pharaohs guarded written knowledge and who could access it. The difference today lies in scale and speed, but the logic of concentrating control over information remains intact.

I don't have all the answers about how reversible this process is. Open models and decentralized approaches offer alternatives, but mass adoption tends to tilt toward those who already command scale. I keep turning this intersection of technology and power over in my mind.

What happens when surveillance is disguised as a useful product, and information curation carries the seal of approval from the very entity that's supposed to guarantee its independence?

Sources:

1. Public reports on negotiations between OpenAI and the U.S. government regarding next-generation models

2. Analysis of layoff and AI infrastructure investment patterns by OpenAI and other companies in the sector

3. Coverage of Meta's expansion strategy for camera-equipped devices