Hurricane Isaias hit the Gulf Coast as a Category 3 storm earlier this month, but a local emergency manager in Alabama used an artificial intelligence chatbot to create a graphic showing where the storm might strike. The image was wrong.
The manager wanted to highlight risks in his county, so he asked the AI to help. The tool did exactly what it was told, yet the meteorological data it produced was inaccurate. Laura Myers, director of the University of Alabama’s Center for Advanced Public Safety, received a call from a meteorologist about the error. Myers notes that while many agencies officially ban employees from using generative AI, the technology is likely already in widespread use within state and local disaster departments.
Why communication fails
Most emergency managers struggle to send clear alerts to the public. This is not a lack of skill but a human limitation. Officials often use jargon, assume people know basic details, and include irrelevant information that distracts from the urgent facts.
Jeannette Sutton, a disaster researcher at the University of Albany, says agencies are limited by resources. She has created a dashboard with templates for disaster alerts. “I think they recognize it as a priority, but they don’t prioritize it until they realize they should have,” Sutton said. “They’re just trying to get their job done without screwing something up.”
At the state and local level, these officials form the first line of defense after extreme weather. The Federal Emergency Management Agency does not handle these immediate responses. Because trust is built through communication, bad messages can destroy that trust.
Flawed human communication has caused problems before. In 2016, adjacent counties in California issued conflicting orders for a wildfire, calling one “voluntary” and the other “mandatory.” In Australia in 2019, officials warned about wildfires using the phrase “Watch and Act.” The government later admitted this caused significant confusion because the public did not know what to do.
Inadequate communication can mean the difference between life and death. During the Thomas Fire in Montecito, California, in 2017, a heavy rainstorm later produced a deadly mudslide. The Santa Barbara County government struggled to warn residents who were still traumatized from the blaze and suffering from warning fatigue.
The county sent alerts to everyone in the “voluntary evacuation zone,” telling them to “stay alert to changing conditions and be prepared to leave immediately.” Most of those residents needed to leave immediately, but the notice sounded optional. Others did not want to evacuate for a second time in a month. The mudslide killed 23 people and injured more than 160 others.
Sutton noted that governments issuing bad warnings lose the opportunity to be the voice that helps people stay safe.
The AI solution
In theory, alerts are a perfect use case for artificial intelligence. Generative AI systems have developed quickly over the past few years. FEMA provides training on synthesizing and sending alerts, but many officials do not remember their training when an emergency arrives.
A large language model can read hundreds of thousands of alerts in minutes, digest research on what works in similar scenarios, and apply that knowledge to edit a human draft in seconds.
“If you’ve only sent one [alert] in your life, and that was 15 years ago, and you need to remember all your training, and you’re thinking ‘Hey, there’s a chemical plume coming my way, the phone’s ringing, the radio’s blaring, I need help,’ the AI is probably gonna do a better job than you are,” said Rob Dale. Dale is a former meteorologist who now runs the emergency management department in Ingham County, Michigan.
Dale has used AI to draft practice alerts behind the scenes but has never had to send one out. Ingham County is a small area in central Michigan that is immune from major weather events like hurricanes and fires. Dale said he would allow a chatbot to draft incident communication if such an event occurred, but he would still want control over sending the final alert.
Capabilities and limits
Some experts believe the technology is ready. EM1, a startup producing AI software for emergency managers, has used Sutton’s research to train a model. Executives say the system is already better than almost any human emergency manager. The company tested this in-house model on scenarios including wildfires and tsunamis. It can match or exceed human capabilities on most benchmark tests.
Sutton gave an example of how the software works using a recent alert from a New Mexico county warning residents about flooding on the San Juan River. The original message read: “Breeched banks. The water is rising. EVACUATE NOW. McGee Park open for people and pets, RV.”
Sutton asked the EM1 model to evaluate and revise the alert. In a few seconds, it produced a far better version that checked all the main boxes of disaster communication.
“San Juan County: Floodwater is rising near [location]. People in [evacuation area]: LEAVE NOW for higher ground. Do not walk or drive through floodwater. McGee Park is open for people and pets. Check [official update source] for updates.”
The model identified and fixed several problems with the original text. The initial message did not say who was sending the alert, name the specific location where flooding was happening, or clarify who needed to evacuate and give advice on doing so safely. It misspelled the word “breached” and did not provide guidance on where residents could seek further information.
EM1 has contracts with around a hundred government agencies and school districts, but none are using it to automate their extreme weather alerts just yet. For the moment, they are only using it to workshop plans and internal preparations. Even so, all the experts interviewed said they are certain that many emergency managers are already using chatbots to write messages, even if they will not admit it.
Sutton noted that the emergency managers who take her alert training sessions often present her with high-quality alerts they have drafted in seconds using ChatGPT or Gemini.
“Official adoption is pretty low, but people are unofficially using it, absolutely,” said Sarah Miller. Miller is a professor of emergency management at Pierce College in Florida and a longtime instructor at the National Disaster Preparedness Training Center. “They’re using their own personal accounts, or they’re using free ones.”
What it means
It sounds like an easy trade-off: if AI is better and faster than humans, and the difference between a good alert and a bad alert can be life and death, we should be using AI for every heat wave and wildfire. Yet even researchers who are bullish on chatbot technology say that making the switch could backfire.
Members of the public who are skeptical of AI or who do not trust the government in the first place may spurn agencies that publish AI-generated graphics or social media posts. In Syracuse, New York, for instance, a union of public employees sued the local government for using an AI service to answer non-emergency 911 calls.
“We trust things that we perceive [are] authentic and care about us,” said Ginny Katz. Katz is the founder of HazAdapt, a consultancy that helps disaster officials figure out how to use technology. “AI-generated content does not feel authentic.”
There is also a risk that chatbots could allow emergency managers to send out more alerts, and faster, which could increase the risk of warning fatigue.
As is the case in other areas where AI is being deployed, disaster experts say that for the moment the software is only as capable as the person using it. A seasoned emergency manager might be able to speed up their communication with the public by having a chatbot draft a series of alerts, but an inexperienced one rushing to create an urgent alert will not know how to evaluate the output of a model like ChatGPT or Claude.
Even though the technology is a ways off from being able to perform without a human expert in the loop, that has not stopped many understaffed emergency management departments from using it. As for how the proliferation of AI-generated content is affecting public trust in government, experts say it is too early to tell.
“If you had a human sending you a text message versus a computer sending you a text message, which one would you believe is being more accurate?” said Justin Snair. Snair is the founder of Preppr, another company that offers AI software to emergency managers.
The answer, he said, depends on the person receiving it.
