U.S. gov't moves to suppress pushback on data centers by removing requirements for public input on pollution — EPA change would allow air pollution permits without publicizing them
A proposed change to the EPA's regulations would remove the requirement for states to seek public input on air pollution permits, particularly around AI data centers.
The U.S. government is contemplating a shift in transparency procedures for data center approvals. The Environmental Protection Agency (EPA) intends to eliminate a requirement that mandates public feedback on applications for air pollution permits, according to The New York Times. This proposed change would affect a subset of air pollution permits, with a particular focus on data centers.
Local communities have been vocal in their opposition to AI data center construction due to concerns over potential environmental and health impacts. The Harvard School of Public Health has estimated that a specific data center in Virginia contributed up to $99 million in annual health damages. In response, 14 states and three cities have signed an open letter opposing the EPA's proposed change, stating that even minor sources of air pollution can have significant health and environmental consequences.
The letter argues that removing the publicization of permit applications and the 30-day comment period undermines transparency and democratic safeguards in the name of administrative efficiency and reducing regulatory burdens. Currently, the EPA requires a public comment period after an air pollution permit application, but in some cases, this process has been bypassed.
For example, an Amazon data center in California began construction without the knowledge of many residents, as the public comment period had already expired. The proposed EPA change would transfer the responsibility for regulating data centers to the states, allowing them to potentially eliminate the public comment period, as it is not a federal requirement.
Written by urgent.news from Tom's Hardware's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.