Author says AI companies are consolidating power at unprecedented levels, monopolizing talent, shaping regulatory frameworks, and cozying up to defense contractors.
Editor’s Note: Karen Hao’s “Empire of AI: Dreams and Nightmares in Sam Altman’s OpenAI” became an instant New York Times bestseller. It has been called “excellent and deeply reported” by Tim Wu in The New York Times and “startling and intensely researched” by Vulture. Others have complained that it is written with a “tendentious slant” and “unwarranted contempt for the technology itself.” Given Hao’s views are being widely distributed, we thought a straightforward overview of the work, with some additional context, would be useful to our readers. However, the opinions expressed are those of the author, not necessarily those of Mogin Law LLP.
Karen Hao, a former AI reporter for MIT Technology Review and The Wall Street Journal, has emerged as one of the artificial intelligence industry’s most incisive critics. In her book, Empire of AI: Dreams and Nightmares in Sam Altman’s OpenAI, and recent interviews on NPR, PBS, and Democracy Now!, Hao sounds the alarm about the AI sector’s trajectory—from idealistic nonprofit roots to a global race for dominance with troubling social, environmental, and ethical consequences.
Hao was the first journalist to profile OpenAI in 2019, embedding with the company for three days. What she found contradicted its public image. Originally founded as a nonprofit to develop AI for the benefit of humanity, OpenAI quickly restructured into a “capped profit” entity, accepting a $1 billion investment from Microsoft. Hao describes this shift as emblematic of Silicon Valley’s “scale-at-all-costs” mentality, where transparency gave way to secrecy and collaboration was replaced by fierce competition.
In interviews, Hao explains that OpenAI’s leadership—particularly CEO Sam Altman—embodies a philosophy of growth and dominance. Altman, she says, is a “once-in-a-generation fundraising and storytelling talent” who can tailor his message to suit any audience. But this talent, Hao argues, has helped mask a deeper ideological clash within the AI community: between those who see AI as a path to utopia and those who fear its existential risks.
The Empire Analogy: AI as Digital Colonialism
The book’s title, Empire of AI, is not metaphorical. Hao draws direct parallels between the AI industry and historical colonial empires. In her view, companies like OpenAI extract labor, data, and natural resources from marginalized communities—particularly in the Global South—while concentrating power and profits in Silicon Valley [3] [4].
She documents how Kenyan workers were paid minimal wages to moderate violent and sexually explicit content for AI training, often suffering psychological trauma [5]. In Chile, Google attempted to build a data center that would consume 1,000 times more freshwater than a local community uses annually—without offering any direct benefit to the residents [3].
Hao argues that this model of development is not just exploitative—it’s anti-democratic. “Democracy cannot survive in a world where the vast majority of people no longer have agency and control over their own lives,” she told NPR[1].
In addition to Hao’s observations, democracy will also struggle if citizens cannot distinguish between, for example, actual video of a president giving a speech and a fake AI-generated video.
Environmental and Political Costs
The environmental toll of AI development is staggering. Hao cites reports showing that data centers may soon consume energy equivalent to multiple Californias, often powered by fossil fuels [3]. These centers also require vast amounts of freshwater, typically drawn from public drinking supplies in water-scarce regions.
While her phrasing that these data centers may consume energy equivalent to “multiple Californias” is rhetorical, research supports the underlying concern. A Cornell University study projects that by 2030, AI-related data centers could emit 24–44 million metric tons of CO₂ annually and require up to 1.1 billion cubic meters of water, largely for cooling systems. These figures equate to the energy use of millions of households and underscore the strain on water resources in drought-prone regions. Such growth, driven by the race to scale AI models, raises urgent questions about sustainability and the industry’s reliance on fossil fuels for power.
Politically, Hao warns that AI companies are consolidating power at unprecedented levels. By monopolizing talent, shaping regulatory frameworks, and cozying up to defense contractors, they are becoming “new forms of empire,” she says [2].
Antitrust and Competition Implications
Market Concentration and Monopolization
As Hao accurately observes, AI development is dominated by a handful of firms, which raises red flags because such concentration limits competition, stifles innovation, and entrenches barriers to entry. Aggressive mergers and acquisitions in AI and cloud computing have already triggered investigations in the U.S., EU, and UK, with authorities scrutinizing deals for anticompetitive effects.
Algorithmic Collusion
As AI systems become autonomous, they can set prices or steer users toward preferred partners without direct human involvement. This creates risks of tacit collusion and discriminatory practices that traditional antitrust frameworks struggle to address. AI price-setting has sparked a considerable number of lawsuits, with courts increasingly ruling against the use of algorithmic pricing tools that thrive when competitors share confidential information. California recently made it clear that algorithmic-facilitated price fixing is still price fixing. Federal bills have also been proposed.
An Opacity Problem
Hao’s critique of secrecy in AI research underscores a collapse of reproducibility, which not only harms scientific integrity but also shields dominant firms from scrutiny. Lack of transparency in datasets and model training makes it harder for antitrust enforcers to assess competitive harm and hold organizations accountable.
The Human Toll
Consumers
Monopolistic control over AI services can lead to self-preferencing and exclusionary practices, limiting consumer options and potentially inflating prices for premium AI tools. Hao also warns that regulatory capture by large corporations could allow these dynamics to persist unchecked. Further, centralized control of data and algorithms erodes consumer agency and can undermine democracy when citizens lose control over their personal data and decision-making processes.
Workers
Hao documents Kenyan and Venezuelan workers earning minimal wages for content moderation and data labeling, often under psychologically harmful conditions. This hidden labor underpins AI’s “magic,” creating ethical and legal risks for companies relying on outsourced work. Jobs themselves are at risk. AI automation threatens employment in sectors exposed to algorithmic substitution, such as telemarketing, data entry and analysis, and bookkeeping and accounting. In the legal world, document review, summarization, and drafting are already being aided by AI. Staff time spent on legal research, client intake, and aspects of marketing are also being reduced thanks to AI. Studies cited by Hao show declines in job security and bargaining power particularly for low- and mid-skill roles.
It’s Not All Bad
Despite her grim assessment, Hao offers glimpses of hope. In the book’s epilogue, she highlights decentralized AI projects like Te Hiku, a Māori language initiative, and the Distributed AI Research Institute founded by Timnit Gebru and Alex Hanna [5]. These efforts aim to democratize AI development and shift power away from tech giants.
Bottom Line
Hao’s warnings point to a convergence of antitrust enforcement challenges, business ethics dilemmas, and social justice concerns. Regulators are already implementing or considering reforms to address algorithmic collusion, merger oversight, and transparency mandates. For businesses, proactive compliance and ethical sourcing are no longer optional. Smaller companies, consumers, and workers should continue to seek remedies in court when AI is used by goliaths to limit, diminish, or erase their opportunities.
References
[1] Journalist Karen Hao discusses her book ‘Empire of AI’
[2] New book ‘Empire of AI’ investigates OpenAI, the company … – PBS
[3] “Empire of AI”: Karen Hao on How AI Is Threatening Democracy & Creating …
[4] Talking with Karen Hao About Empire of AI and the Colonizing Logic …