Emergency managers warn AI-generated disaster alerts risk inaccuracies and public distrust

Unofficial AI adoption in emergency alerts

Emergency management agencies are grappling with the unofficial use of generative AI tools, even as official policies lag behind. Laura Myers, director of the University of Alabama’s Center for Advanced Public Safety, noted that while most agencies do not formally permit AI usage, its adoption is likely already pervasive.

One instance highlighted the risks: a local emergency manager in Alabama used an AI chatbot during Hurricane Isaias to generate a graphic predicting a "sting jet" of high winds. The graphic, however, was meteorologically inaccurate, underscoring the potential for AI to spread misinformation.

Jeannette Sutton, a disaster researcher at the University of Albany, warned that AI-generated alerts may not feel authentic to the public, potentially eroding trust in critical communications.

AI tools marketed to emergency managers

Hundreds of companies are targeting emergency management agencies with AI tools designed for tasks like flood damage assessment and alert drafting. EM1, a startup, claims its AI model is already better than most human emergency managers at drafting alerts. The company has contracts with around a hundred government agencies and school districts, though it is not yet used to automate extreme weather alerts.

Despite these claims, experts remain cautious. Rob Dale, head of emergency management in Ingham County, Michigan, emphasized that public trust in government alerts is fragile. Unclear or conflicting messages, whether AI-generated or human-written, can have deadly consequences, as seen during the 2017 Montecito mudslide, where poor communication contributed to 23 deaths.

Challenges in human and AI-generated alerts

Human-written alerts often suffer from jargon, omitted information, and irrelevant details, making them difficult for the public to understand. Sarah Miller, a professor of emergency management at Pierce College, pointed out that emergency managers frequently struggle with clarity, which AI tools could potentially address.

However, the effectiveness of AI-generated alerts depends on the recipient’s perception of their accuracy and trustworthiness. Ginny Katz, founder of HazAdapt, and Justin Snair, founder of Preppr, both acknowledged that while AI can improve efficiency, it may also exacerbate warning fatigue if alerts lack authenticity or relevance.

What this means

Lazyfounder analysis — our interpretation, not reported fact.

The rise of AI in emergency management reflects a broader trend in GovTech: the pressure to adopt tools that promise efficiency, even when official policies and public trust haven’t caught up. For founders and operators in this space, the lesson is clear—AI can augment human decision-making, but its deployment requires rigorous validation to avoid misinformation and maintain credibility. The unofficial use of AI tools also signals an urgent need for standardized guidelines, lest agencies risk reputational damage or, worse, public harm.

Key takeaways

• Founders in govtech and emergency management must prioritize accuracy and transparency in AI tools to avoid eroding public trust.
• The unofficial adoption of AI in emergency communications signals a need for clear guidelines and oversight to prevent misinformation.
• Startups like EM1 demonstrate the potential for AI to improve alert drafting, but human oversight remains essential to ensure reliability.
• Warning fatigue and public skepticism could intensify if AI-generated alerts lack clarity or authenticity, making trust a critical factor in adoption.

Related on Lazyfounder

• More ai news