Book Review — “Empire of AI” by Karen Hao

When I first cracked open Karen Hao’s Empire of AI: Dreams and Nightmares in Sam Altman’s OpenAI, I had high hopes for it. Here was an acclaimed tech journalist, formerly the senior AI editor at MIT Technology Review and a foreign correspondent for The Wall Street Journal, with a front-row seat to the largest artificial intelligence story of our generation. The prologue, rich with behind-the-scenes detail on Altman’s brief ousting and dramatic reinstatement in 2023, grabbed me immediately. It promised the kind of reporting that only someone with deep sources and years of coverage could deliver. I was already salivating that this is Steve Coll-level stuff.

By the time I reached page 350, however, I found myself both better informed and deeply irritated. I dumped it.

Hao’s book, published in May 2025 and running a dense 450 pages, is marketed as the definitive inside account of how OpenAI, the company behind ChatGPT, morphed from an idealistic non-profit co-founded by Elon Musk to an empire with Sam Altman at its manipulative, messianic center. There’s no question she’s done her homework: the book draws on more than 300 interviews with current and former employees, insiders at Microsoft and Google, and a library of leaked emails and Slack messages.

At its best, Empire of AI is a brisk, alarming look at how powerful Silicon Valley players, from Altman to Musk backstabbed, and rebranded their moral crusade for “safe” AI into a for-profit race that pushed any notion of openness out the window. Hao does a solid job showing how Altman, once a footnote in Elon Musk’s circle, outlasted Musk, and has now charmed the president from recent news.

Yet for all this strong factual ground, Empire of AI never quite settles into a balanced take. It’s not that Hao is critical, good journalism often is, but that her book pivots so quickly from credible reporting to ideological soapboxing that it’s hard to take some parts seriously. Her obsession with positioning OpenAI as a new colonial empire, devouring data and cheap labor from the Global South, is not unfounded, but the point gets hammered so relentlessly that it drowns out more original insights.

I lost count of how many times she rehashes the same metaphor: “AI empires are the new colonial powers.” She’s not wrong to highlight the miserable wages of Kenyan data labelers or the energy-hogging data centers in Arizona and Chile. But when every chapter circles back to this moral indictment, with entire pages pivoting from GPU clusters to vague laments about global inequality, you begin to suspect the author is more interested in scoring political points than unpacking what AI actually is or how it works.

Therein lies the book’s core problem: Hao has no love for AI, nor much respect for the people building it. In one stretch, she appears genuinely annoyed that the term “artificial intelligence” doesn’t even have a clear definition, a real irony for a 450-page book about it. While she’s quick to report insider gossip or Altman’s strategic mind games, she rarely explains how the actual technology functions or where the science might genuinely transform industries for better or worse. The result is a lopsided critique: part fact, part blog rant.

One recurring theme that grows especially grating is Hao’s tendency to wedge her real target, the patriarchy, big tech’s racism, and systemic bias, into places where it sometimes feels shoehorned. AI’s structural harms do deserve scrutiny, but at points her writing slides into a broad polemic about women and black people in tech with only loose ties to the book’s core story. It feels like you’re reading a social science thesis disguised as a corporate biography.

This would be easier to forgive if Hao showed more insight into how AI might actually deliver on its promise, or fail. But time and again, she offers no clear technical counterpoints. Instead, we get predictable doom scenarios and what-ifs about job loss and surveillance, peppered with shallow asides about “empire logic.”

It’s frustrating because when Hao stays on track, she’s genuinely good. Her portrait of the OpenAI meltdown like when Altman was abruptly fired by his board, only to return days later with Microsoft’s muscle behind him is clear-eyed. She lays out how a nonprofit board designed to safeguard humanity from rogue AGI ended up torn apart by secrets and ego. The behind-the-scenes detail, the legal maneuvering, the panic among staff are the gold standard of tech reporting.

Likewise, the early chapters deliver a fast-paced history lesson on how OpenAI’s co-founders pitched their moonshot as a check against Google’s DeepMind, only to follow the same path of secrecy and aggressive scale. We learn how Musk’s paranoia about Demis Hassabis fueled OpenAI’s origin myth, and how Altman turned the lab into a $150 billion empire with a quasi-religious mission statement.

Yet by the final third, it’s clear that Hao would rather scold than probe. There’s no meaningful attempt to unpack the science that fuels models like GPT-4. There’s no real exploration of how generative AI could be deployed responsibly, or why so many ordinary people find these tools useful — even joyful. Where Noah Harari’s Nexus (a book I’d recommend over this) at least tackles these questions with a philosophical lens, Hao’s Empire of AI defaults to gloom and scorn.

Should you read it? That depends. If you’re new to the AI circus and want a crash course on the people behind the headlines like the chessboard politics and the venture capital backroom deals, there’s value here. But if you want a book that helps you understand AI’s science or practical implications for business, you’ll find Empire of AI shallow and repetitive.

My verdict is that the book is a cautionary story worth skimming for its corporate intrigue. But if you want real insight into the future of artificial intelligence, pick up something else abeg.