Karen Hao Frames AI as Threat to Democracy

In a Democracy Now! interview segment - part of the network's July 3, 2026 July Fourth special revisiting a 2025 conversation - journalist Karen Hao, author of Empire of AI: Dreams and Nightmares in Sam Altman's OpenAI, told hosts Amy Goodman and Juan Gonzalez that Silicon Valley's AI buildout mirrors historical colonial extraction: "The empires of AI ... seize and extract precious resources to feed their vision of artificial intelligence: the work of artists and writers; the data of countless individuals ... the land, energy, and water required to house and run massive data centers and supercomputers," she writes in the book. Hao, a former Wall Street Journal and MIT Technology Review reporter who leads the Pulitzer Center's AI Spotlight Series for journalists covering AI, pointed to her reporting from Kenya, Chile and Uruguay as evidence.

On energy, Hao cites a McKinsey report projecting that AI data center expansion could add as much demand to the global grid within five years as two to six times California's annual electricity consumption, mostly met by fossil fuels, coal-plant life extensions and unlicensed methane turbines. On water, she cites a Bloomberg analysis finding two-thirds of new data centers are sited in water-scarce regions. Her most detailed example: a Chilean community near Santiago that still holds a rare public freshwater right pushed back after Google proposed a data center there that would have drawn roughly 1,000 times the community's annual water use; sustained local activism escalated to Google Chile, Google's Mountain View headquarters and eventually the Chilean government, blocking the project for four to five years and winning a seat at a new roundtable process. On labor, Hao's reporting describes Kenyan contract workers hired by data-annotation firms to label OpenAI's most graphic AI-generated content for a content-moderation filter, paid a few dollars an hour, which she contrasts with the compensation of Silicon Valley-based AI researchers.

Hao also flagged a provision in the House-passed version of the 2025 U.S. budget reconciliation bill that would have barred state governments from regulating AI for ten years, opposed even by some Republican supporters of the broader bill, including Rep. Marjorie Taylor Greene, who said she had not known the provision was included when she voted for it. The Senate ultimately stripped the AI moratorium from the bill in a near-unanimous 99-1 vote on July 1, 2025, so it did not become law - though Hao's broader argument, that major AI developers are actively working to insulate themselves from state-level oversight, remains a live debate, since similar federal preemption proposals have continued to surface since.

Teams building or procuring AI infrastructure should treat community water and energy impact as a live commercial and reputational risk, not a hypothetical one: the Chilean dispute shows a single contested data center site can be delayed for years. The Kenya example is a reminder that content-moderation and data-labeling supply chains carry labor and psychological-harm exposure that traces back to the model owner even when the work is subcontracted. Organizations tracking U.S. AI policy should note that federal preemption of state AI law was defeated in 2025 but remains a recurring legislative proposal, not a settled question.

Whether federal AI-preemption language resurfaces in future legislation, following the pattern of the 2025 attempt; further reporting on data center siting in water-stressed regions; and whether AI labs disclose more about content-moderation labor conditions and compensation.

Hao's "empire" and "colonialism" framing is her own interpretive argument built from the specific reporting above, not a claim about any individual executive's private intent; the underlying facts - the McKinsey and Bloomberg figures, the Chile dispute, the Kenya labor arrangement, and the state-preemption bill's fate - are independently attributable and, where checked, confirmed.