Just in time for hurricane season and what promises to be a summer of record-breaking heat, the Trump administration has decided that the best way to improve weather forecasting is to slashing the data that makes forecasting possible. Experts are calling this a bold, and possibly catastrophic, strategy.
The National Oceanic and Atmospheric Administration (Noaa) launched a suite of AI-powered global weather forecast models late last year, boasting improvements in “speed, efficiency, and accuracy.” By March, those models were being trained on centuries of weather data. The catch? AI is only as good as the data it’s fed, and under Trump, climate and weather data collection has taken a nosedive. Monica Medina, Noaa’s former principal deputy undersecretary of commerce for oceans and atmosphere, notes the agency faces a proposed 40% cut overall, despite a modest budget increase for the National Weather Service. “We absolutely need AI to help us crunch the data faster,” she says. “But right now, what we’re doing is cutting back the data collection … we’re going in the wrong direction.”
Noaa spokesperson Erica Grow Cei insists there’s “a wealth of weather data collected each day,” but widespread reports tell a different story: staffing cuts have forced the agency to scale back satellite launches, balloon launches, and ocean buoy networks. Craig McLean, Noaa’s former acting chief scientist, puts it succinctly: “Weather times time equals climate. Cutting climate research impacts the skill of our weather forecast, and it arrests our advancement of weather forecasts.” All this as a “super El Niño” threatens to spike temperatures and boost hurricane activity.
To make matters more interesting, the AI models Noaa is leaning on have a known weakness: they underperform during extreme weather events, according to a study published in Science Advances. Because they’re trained on historical data, they struggle to predict the record-breaking events that are increasingly common in a warming world. Traditional physics-based models don’t have this problem, because they don’t care what happened before - they just follow the rules. “The AI weather models were trained on a climate that no longer exists,” says forensic meteorologist Chris Gloninger, who received death threats for discussing climate change on TV. He notes that conventional models outperformed AI-based ones during a historic blizzard in February 2026.
Noaa administrator Neil Jacobs, a Trump appointee, defended the cuts at a House subcommittee hearing in April. Former colleague John Sokich believes Jacobs wouldn’t rush to implement untested AI, but McLean worries that “the man has demonstrated his willingness to be obedient to the president who appointed him [and who is] destroying the National Oceanic and Atmospheric Administration.” Less accurate forecasts could mean more dangerous outcomes for Americans. “Weather forecasts are vital to our economy, to our health, and to public safety,” Medina warns. So as hurricane season looms, the country may be about to learn just how much we rely on data we’re now choosing not to collect.