On 17 November 2023, Sam Altman was fired from OpenAI, the company he had helped found, shape, and turn into a $90 billion behemoth. The decision was announced over a Google Meet, and it came, as so many of these things do, just as he was preparing to attend a Formula One party in Las Vegas. For a moment, the AI revolution appeared to have hit turbulence. Within five days he was back at the helm, the board reshuffled, his myth burnished. The lesson, depending on who you ask, was either that the old order had held—the founder had fallen—or that the new order had prevailed—the founder was irreplaceable. For Sam Altman, it was Tuesday.
That week serves as the narrative hinge of two books published on the very same day of each other: Empire of AI by Karen Hao, and The Optimist by Keach Hagey. The books share a cast. Altman, Musk, Thiel, Sutskever. And a plot: how a non-profit dedicated to building safe artificial intelligence became the most valuable startup in the world, the crown jewel in Microsoft’s imperial ambitions. But the similarity ends there. Hagey sketches a character, mostly in admiration. Hao, scalpel in hand, performs an autopsy. Hao, a former MIT Tech Review journalist, has written what is likely to become the foundational critical history of AI’s new industrial phase. Hagey, a Wall Street Journal reporter, has written a biography of its high priest.
The title of Hao’s book is not metaphorical. She means “empire” in the classic Marxist sense: an arrangement of production, extraction, ideology and violence, stretched across borders and rationalised by modernity. In her telling, AI is not simply a neutral tool, but a world-making apparatus built on colonial logics. The training data is scraped from writers, artists and users without consent; the labour is outsourced to precarious workers in the Global South; the electricity and water are siphoned from resource-poor municipalities. The industry calls this “alignment”. The rest of us might call it theft.
“It was never just one company,” she writes. “It was a worldview.”
The Optimist never quite comes to this conclusion. Hagey is a careful writer and a fair one, and she has done more than 250 interviews to document Altman’s rise, from awkward gay teenager in St Louis to techno-visionary promising fusion energy, crypto-driven UBI and brain implants. But she cannot quite bring herself to question the foundational tenets of her subject’s worldview. She writes, admiringly, that Altman “doesn’t just want to create a new technology and give it to the world. He has always aspired to be… a ‘great man of history’.” At the very moment when OpenAI’s employees, governance structure, and mission collapsed under that ambition.
OpenAI’s origin story begins, as all Silicon Valley myths do, with a group of earnest men declaring they would not be like the others. Elon Musk, Sam Altman, Ilya Sutskever and a handful of others launched OpenAI in 2015 as a non-profit committed to developing “safe” artificial general intelligence—AGI—not for shareholder profit but for the benefit of humanity. There would be no race to the bottom, no secret patents, no Google-style walled garden. Their research would be open. If someone else got there first, they would stop and help. It was less a start-up than a secular monastery, solemnly pledged to steward the singularity.
By 2019, those vows had been quietly set aside. Musk had left in a huff; Altman, having outmanoeuvred him for power, reincorporated the venture as a for-profit capped at 100x returns. Microsoft arrived with the cheque book. The company stopped publishing its models. Then came ChatGPT, the fastest growing consumer app in history, followed by the board coup, the investor revolt, the reconciliation. Safety became a slogan, scale became the goal.
Both books agree on the facts, but not on their meaning. Hagey sees the pivot from safety to commercial dominance as a regrettable but necessary compromise. Hao sees it as a lie. The safety discourse was always useful. Both as internal cohesion and as external legitimation. But it obscured the deeper ideology: scale-as-progress, expansion-as-ethics, growth as both solution and reward. The safetyists, mostly young Effective Altruists drawn from LessWrong forums and Future of Humanity institutes, eventually lost the argument. They were outmanoeuvred not by malice but by the logic of the machine: investment demands returns, scale demands secrecy, product demands pace. In November 2023, when board member Helen Toner suggested the company’s actions might still be justified even if they destroyed it. So long as the mission was preserved. She was not being rhetorical. She had misunderstood the mission.
If Hagey frames OpenAI as an outgrowth of Altman’s psyche. His messianism, his anxieties, his “add a zero” philosophy. Hao zooms out. Her argument is that the ideology of scale is not just a personal fixation, but a new imperial mode. Empire of AI positions OpenAI not as a company, but as a node in a global infrastructure of plunder. She traces the training data back to the artists and users who never gave consent. She follows the machine-learning pipeline down to Kenya and the Philippines, where low-paid workers tag violent content and racist slurs for pennies. She investigates the environmental cost. Arizona’s water, Chile’s energy, lithium for data centre batteries and finds that the “cloud” is not so ethereal after all.
These are not metaphors. Hao is making a materialist claim. AI systems don’t just require data, they devour it. They don’t just need labour, they exploit it. They don’t just use electricity, they consume it at industrial scale. OpenAI’s GPT-4, she notes, is more than 15,000 times larger than GPT-1, and trained on a supercomputer that runs hotter than most cities. The fantasy that AI floats above the world dissolves quickly on contact with reality. “Labs”, “models”, “clouds”—these are not weightless abstractions. They are built by human hands, using stolen words, scraped images, and planetary resources, for the enrichment of a few.
In this, the empire metaphor becomes less an analogy than a diagnosis of the current. The AI boom mirrors the colonial one: extract resources from the periphery, refine them in the core, export power back outward. As with empire, the moral justification is always civilisation. OpenAI’s charter promises global benefit. Its executives talk of tutoring every child, curing cancer, solving fusion. But these promises come wrapped in asymmetries: the harms are immediate, distributed, and invisible; the benefits are speculative, concentrated, and always just over the horizon.
The question that haunts both books is the same: who is Sam Altman? Hagey tries to answer it in the traditional biographical sense—childhood, mentors, relationships, investments. We learn that he is gay, grew up in St. Louis, came out on AOL Instant Messenger, and was a precocious programmer who turned his first app into a start-up before he turned twenty. We are told he is affable, reticent, anxious, wildly ambitious, a skilled networker and a terrible manager. The line she repeats, from Paul Graham, is instructive: “You could parachute him into an island full of cannibals and come back in five years and he’d be the king.” That is meant as a compliment.
Hagey’s Altman is a paradox: personally modest, institutionally imperial. He doesn’t crave money but amasses it anyway; he claims not to seek control but always ends up in charge. His rise is explained in the Silicon Valley idiom of “vision” and “execution”, as though the world bends for those who simply believe hard enough. The idea that belief itself might be a form of power and that Altman’s power is not merely the product of charisma but of structural forces, does not detain the narrative long. To ask why Altman commands such influence is, for Hagey, to ask why Copernicus moved the sun. He just did.
Hao is not interested in Altman as a personality. Her portrait is colder, sharper, and ultimately more damning. Altman is not an individual but a type: the founder-visionary, the scale-evangelist, the soft-spoken prophet of technological salvation. She is interested in what he represents, not what he feels. His shift from “safety” to “sovereignty”, from collaborative research to monopoly control, is treated not as betrayal but culmination. He didn’t betray OpenAI’s mission; he fulfilled it.
The prologue of Empire of AI recounts the November 2023 firing not from Altman’s point of view, but from the panic of workers watching their equity vanish, the silence of leadership, the meltdown of governance. It is not a tragedy, it’s a coup. The details are granular. Slack messages, Signal threads, midnight meetings and the conclusion is stark: OpenAI’s structure was always a façade. Altman had designed it so he could be removed, but only by those who never would. When they tried, they were replaced. “OpenAI is nothing without its people,” went the employees’ open letter. The truth, as Hao shows, is that OpenAI is nothing without Sam.
What emerges across both texts is a modern fable of rule. Altman, like the tech titans before him, has become the vector through which empire dreams of itself: benevolent, brilliant, burdened. Hagey calls him an “optimist”; Hao calls him an emperor. They are both right. The former believes history bends toward abundance; the latter knows who gets dispossessed along the way.
At the heart of both books lies a belief. Stated plainly in one, scrutinised in the other. That artificial general intelligence will change everything. It will remake work, economy, statecraft, war. Altman’s own vision, repeated in countless interviews, is that AGI will be “an extension of our wills”, a tool so powerful it renders scarcity obsolete. Universal basic income will follow. Education will flourish. Cancer will be cured. We will, as the line goes, have abundance.
But whose abundance? Hao reminds us that AGI is not a metaphysical inevitability. It is a product, built by workers, using extractive inputs, financed by venture capital. There is nothing neutral about it. What OpenAI calls “alignment” is a vision of the future structured around their image of reason, risk, and reward. The models are trained on our words, our lives, our labour; the profits accrue elsewhere. The logic of AGI-as-salvation is not so different from other imperial logics: it promises transformation, and demands obedience.
It’s also worth noting how AGI is increasingly spoken of in theological terms. Altman himself has said that the most successful people “create religions1”. His language is saturated with revelation: “pulling back the veil”, “the frontier of discovery”, “deliverance”. In The Optimist, this is taken at face value. As a kind of modern mysticism of code. In Empire of AI, it is a warning. For Hao, the myth of AGI is not just a dream of better tools. It is a utopianism that obscures how power is already being redistributed upward, under cover of innovation.
The consolidation is happening in plain sight. Microsoft now has privileged access to OpenAI’s systems. Google has folded its AI divisions into a centralised DeepMind. Anthropic is flush with Amazon cash. Nvidia, the chip supplier, is the bottleneck of it all. The startups orbit the giants. The researchers go quiet. Alignment, in practice, means centralisation: fewer voices, fewer paths, more control.
Hao understands something Hagey can’t, or won’t, quite say, is that the future imagined by OpenAI is not inevitable, nor neutral, nor universally beneficial. It is a future with a shape, and that shape has a billionaire owner. It always does.
Footnotes
Alexander Dugin (11) Artificial Intelligence (11) Book Review (93) Books (97) Britain (66) Capitalism (10) Conservative Government (36) Creeping Fascism (12) Crime and Punishment (14) diary (11) Donald J Trump (64) Europe (12) Far Right Extremism (11) Fascism (10) Film (13) France (16) Gaza (21) History (12) Imperialism (23) Iran (10) Israel (23) Keir Starmer (14) Labour Government (46) Labour Party (10) Migrants (20) Nigel Farage (21) Palestine (17) Protest (18) Reform UK (31) Russia (23) Television (10) Ukraine (12) United States of America (106) War (26) Work (11)
