Empire of AI

( em português: sol2070.in/2026/02/empire-of-ai-karen-hao/ )

Empire of AI (2025, 496 pp.) is a meticulous deep dive into OpenAI (of ChatGPT) and the actions of its leaders, by reporter Karen Hao, perhaps the journalist who has covered the corporation most extensively.

As the title suggests, the angle is the creation of a techno-empire, or how new technologies reinforce structures of that kind.

Hao says in the introduction:

While it tells the inside story of OpenAI, that story is meant to be a prism through which to see far beyond this one company. It is a profile of a scientific ambition turned into an aggressive ideological, money-fueled quest; an examination of its multifaceted and expansive footprint; a meditation on power.

The book really goes deep into how technology companies work and, indirectly, how the world works.

In addition to being an impressive piece of reporting for its depth—especially considering it was not officially authorized or supported—the narrative grips almost like a thriller.

It revolves heavily around the dissimulation of leader Sam Altman, whose behind-the-scenes behavior is very different from his public-relations good-guy image (for example, while OpenAI was receiving billions and all the spotlight, his sick sister, with whom he refused contact, had to enter the sex industry to support herself).

Colonial tactics become very clear in the chapters on the exploitation of poor people in countries such as Kenya and Colombia.

Since the mass of data on which ChatGPT was trained includes the worst the internet has to offer, the task of sanitizing it, as always, fell on multitudes of people in misery earning pennies for unhealthy work. To do this, they were practically held hostage 24 hours a day by an uberized platform that distributes content-classification tasks (which include pedophilia, bestiality, decapitation, etc.).

Obviously, when questioned, the corporations that benefit from this exploitation deny everything, blaming the companies they contract.

A constant theme is AGI (artificial general intelligence), the term for a dreamed-of AI with the same human capacities that could also rapidly self-improve toward a superintelligence.

So far, this is just speculation—but highly profitable speculation. Companies in the field pursue it in such a believing, fanatical way that it has already turned into a techno-religion.

Altman professes total belief in AGI, but, as always, it is doubtful how honest he is. He is probably also manipulating expectations for his own benefit.

As he himself said at the time when he worked at Silicon Valley’s master incubator, Y Combinator (according to this excerpt from the book):

“The most successful founders do not set out to create companies,” Altman reflected on his blog in 2013. “They are on a mission to create something closer to a religion, and at some point it turns out that forming a company is the easiest way to do so.”

To step away a bit from techno-imperial oppression, there is an epilogue at the end about how “AI” could be used to truly benefit humanity. One detailed example is the language model developed for the recovery and revitalization of a native Māori language in New Zealand. The people involved did everything independently, open-source, far from corporations, with exemplary success.

As one researcher put it, there are more precise ways to specify what would actually constitute the “benefit to humanity” proclaimed by every techno-billionaire:

… during the Queer in AI workshop, Ria Kalluri, an AI researcher at Stanford, proposed an incisive alternative to the question of how to ensure AI does “good.” Goodness, benefit to humanity—these terms will always be in the eye of the beholder. Rather, we should ask how AI shifts power: Does it consolidate or redistribute that power? To put it in the frame of this book, does it continue to fortify the empire, or does it begin to wrest us back toward democracy?

In Empire of AI, Karen Hao has built a compelling portrait of the internal workings OpenAI, and through it, a portrait of both the AI movement and the modern tech industry. I hope that everyone who has contemplated bringing AI into their businesses and workflows reads it: the tools and even the promise of the technology has been shaped by these specific people and their assumptions, values, and worldviews.

The title is apt. OpenAI’s approach is empire — this is an organization that seeks to centralize power rather than democratize it. It readily sits on a shelf alongside books like Careless People, but the themes here are sometimes reminiscent of Succession. It’s about power. The whole industry is about power. And Hao lays it out clearly and beautifully.

We’re left with a little sprig of hope in the epilogue. I agree with the premise here: there’s a version of these technologies that are tightly bound, built respectfully, and aim to do one thing well. That’s the future. This idea that there should be one tool that does everything well, built with many billions of dollars, the exploitation of the global majority, and with neurotic executives with egos the size of warships at the helm, is not long for this world. May it depart quickly and with as little further damage as possible in its wake.