Despite these impacts, the problem appears overlooked by the tech industry. "If you examine tech companies' sustainability reports, they mention carbon emissions (碳排放) and water use, but completely ignore harmful air pollutants already damaging public health," explains Shaolei Ren, a study author.
The research team recommends that tech companies report air pollution from their electricity use and generators. They also call for proper compensation for communities suffering the worst health effects from this pollution.
Notably, AI-related air pollution harms certain low-income neighborhoods more severely, partly because they often sit closer to power sources. Furthermore, pollution spreads across state boundaries. "Data centers pay local taxes," Ren notes, "but health impacts reach nationwide. Other affected areas receive no compensation." For example, pollution from Virginia data centers spreads into seven neighboring regions.
As tech companies compete to launch new AI services, dangerous air pollutants — especially tiny particles entering lungs — are expected to rise sharply. By 2030, health damage from AI pollution could match that of all vehicles in California.
"AI's growth hugely increases demand for data centers and energy, making it the fastest-growing energy user across all industries," Ren states. Training just one major AI model creates pollution equal to 10,000 car trips between Los Angeles and New York.
"If your relatives have asthma or heart conditions, this pollution may harm them today. This serious health problem cannot wait," Ren warns.