Karen Hao's Empire of AI is a deeply researched investigation into the rise of OpenAI and, by extension, the political, technological, and social stakes embedded in contemporary AI development.
Drawing on hundreds of interviews and a significant body of internal correspondence and documentation, Hao reconstructs the organizational transformation of OpenAI from a nonprofit research collective to a well-funded, secretive corporate actor at the center of what she calls a new imperialism of artificial intelligence.
The book excels in three interrelated domains. First, as a study of OpenAI as an industry leader, it offers granular insight into the internal operations, shifting ethos, and strategic realignments of OpenAI from its founding through the ChatGPT era.
Second, as a psychological portrait, it casts OpenAI as an expression of Sam Altman's personality, whose oscillation between techno-idealism and market opportunism mirrors the broader contradictions of Silicon Valley.
Third, and most critically, Hao draws upon the history of imperialism and colonialism to situate OpenAI's trajectory within a global frame: as the vanguard of an emerging techno-empire that extracts labor, data, and ecological resources from a planetary periphery to build models whose risks and benefits are managed by a small, remote, powerful billionaire elite.
Listen to Karen Hao’s Empire of AI at Audible
Hao brings to this analysis a rare breadth and depth of understanding of the intersection between technology and society. Trained as an MIT scientist and experienced as a tech journalist, she writes with clarity and precision about both machine learning architectures and their political implications.
Most notably, she advances a historical analogy between AI development and colonialism. The comparison is not rhetorical flourish but a structural argument: the concentration of technological power, the subordination of marginalized labor, and the appropriation of common resources serve as pillars of both systems. Hao does not merely diagnose this dynamic but offers an alternative: a discrete, labor-enhancing vision of AI that supports rather than supplants human capacities.
The book’s empirical detail and narrative structure are compelling, but its normative ambition is what sets it apart. Hao does not assume technological determinism. Rather, she insists that AI systems, like empires, are built through political choices. Her critique of OpenAI is not reducible to personal antagonism or institutional mismanagement. It is a claim about the incompatibility of democratic accountability with the current mode of AI production.
The Audible edition, which Hao narrates herself, reinforces this message with precision and conviction, her delivery augmenting the book’s clarity and urgency of purpose. Empire of AI thus contributes not only to public understanding of OpenAI but to a wider reckoning with the infrastructures of digital power.
Hao's book is not just an exposé of a single firm but a field report from a contested future. It is valuable not because it answers all questions, but because it frames the right ones: Who builds AI? For whom? With what resources? And at what cost? The book has provoked fierce opposition, most notably from OpenAI's leadership.
Sam Altman has publicly criticized Hao's portrayal of the company and its mission, while other figures within the AI industry have also sought to discredit her reporting. These responses underscore the stakes of Hao's challenge to not only technological practices but the authority of those who currently shape the agenda. This makes the book all the more significant and well worth reading.