Empire of AI
by Karen Hao · 2025
An investigative journalist's inside account of OpenAI, tracing the gap between its nonprofit-for-humanity founding story and what it actually became.
Worth reading? Hao covered OpenAI as a journalist for years before writing this, and it shows in the access and the skepticism: she treats 'building AI for the benefit of humanity' as a claim to investigate, not a mission statement to repeat. The book traces the founding, the Musk fallout, the Microsoft deal, and the November 2023 board crisis as evidence of a company whose structure kept bending toward speed and scale over its original safety-first charter. Read it if you want the reporting behind the headlines. Skip it if you're after AI capability content, not corporate history.
| Full Title | Empire of AI: Dreams and Nightmares in Sam Altman's OpenAI |
|---|---|
| Author | Karen Hao |
| Published | 2025 |
| Category | Business & Money |
The Verdict
Hao isn’t writing a hit piece or a hagiography, she’s doing the reporting most AI coverage skips: what actually happened inside the building when the mission statement met the cap table. If you’ve only followed OpenAI through press releases, this is the corrective.
readers who want the reporting on how OpenAI's mission and its business reality diverged
you want technical AI content rather than corporate/organizational history
Book Summary
Hao's throughline is the tension baked into OpenAI's structure from day one: a nonprofit mission ('ensure AGI benefits all of humanity') wrapped around what became a capped-profit company racing competitors for capital and compute. She argues that tension, not any single bad actor, explains the recurring pattern of safety commitments getting renegotiated whenever they collided with growth.
She frames OpenAI as the leading example of an 'empire' model, a small set of companies and their backers effectively deciding the terms of a technology that will reshape labor, information, and power globally, with far less democratic input than the stakes would suggest. The 2023 board ouster and reinstatement of Sam Altman is presented as the moment that tension became visible to the public.
Top 7 Lessons from Empire of AI
- A nonprofit mission statement doesn't guarantee nonprofit-style decision-making once venture capital is involved.
- Structural incentives (compute costs, investor pressure, competition) shape safety choices more than stated values do.
- The 2023 OpenAI board crisis exposed a governance structure unable to enforce its own founding charter.
- Companies calling themselves 'labs' still make decisions under the same commercial pressures as any other business.
- Access to frontier AI is concentrating in a handful of well-capitalized firms, with limited public oversight.
- The gap between a company's public mission and its internal incentives is worth investigating, not assuming away.
- Speed-to-market and safety-first commitments are frequently in direct, unresolved tension.
Frequently Asked Questions
Is Empire of AI worth reading?
Yes, if you want investigative reporting on how OpenAI's founding mission collided with its commercial reality, especially around the 2023 board crisis.
What is Empire of AI about?
Journalist Karen Hao's account of OpenAI's evolution from a safety-focused nonprofit to a capped-profit company racing for AI dominance, and the tensions that created.
Who should read Empire of AI?
Readers who want the organizational and journalistic story behind AI headlines, not technical AI content.
Ready to read it?
Get Empire of AI on Amazon
