How AI agents can help emergency teams act before a crisis escalates

akinbostanci via Getty Images
COMMENTARY | Agents can identify new signals, test them and surface an assessment for human review, meaning they can contextualize threats quicker so teams can assess risk in real time.
State and local cyber and emergency response teams have more information than ever. What they often lack is enough time and situational context to determine what matters so they can act before an incident escalates.
Autonomous agents run on artificial intelligence can help close that gap by continuously evaluating emerging signals, comparing external threat signals with internal security telemetry and updating assessments as conditions change. These systems can give response teams earlier insight into what is happening and where to focus first.
By the time a state or local team has manually gathered enough information to confirm an event, the opportunity to limit the damage may already be gone. The volume and velocity of information, communications and public reporting have outpaced what human-led monitoring can evaluate in real time.
AI agents can identify new public signals, test them against an organization’s internal security telemetry and surface an assessment for human review. As new information appears, they can reevaluate their conclusions without waiting for an analyst to restart the process.
The result is that these agents can contextualize external threat signals in a way that would take days for humans and present it in real-time so cyber teams can assess risk and develop mitigation strategies.
Understanding the Stakes
Anticipating cyber threats has become a critical need in defending state and local networks and systems.
The FBI's Internet Crime Complaint Center tracked more than 3,600 ransomware complaints in 2025, including more than 2,100 involving U.S. critical infrastructure. The Cybersecurity and Infrastructure Security Agency has also reported that the Medusa ransomware operation compromised more than 500 organizations in U.S. critical infrastructure sectors after emerging in 2021. These numbers prove that despite the steps they’ve already taken, state and local organizations still face significant gaps in the time between spotting risk indicators around vulnerabilities and exploitation.
State and local governments are directly exposed. They operate essential infrastructure and public services, often with limited cybersecurity resources. At the same time, AI is increasing the speed at which adversaries can identify weaknesses and carry out attacks.
A cyber incident can also quickly become an operational emergency. A disruption affecting transportation or another public service can require decisions that go well beyond the security operations center and into the territory of public safety. That makes the speed and quality of the underlying intelligence and communication important across the broader response organization.
Agentic Corroboration in Action
Traditional monitoring often requires an analyst to receive an alert, investigate it, search for related information and determine whether the signal is credible enough to escalate.
An autonomous agent can begin those steps as soon as a signal appears. It can continuously compare public reporting with cyber telemetry, partner alerts and operational data. It can also incorporate emergency communications and infrastructure information as circumstances evolve. When evidence supports a connection between events, the agent can surface that relationship and explain the basis for its assessment.
Consider a cyberattack against critical transportation infrastructure. Threat actors often advertise an attack before a targeted agency has even confirmed a breach occurred, and that gap is where early, autonomous detection creates value.
One example comes from an April 2026 incident involving public transit infrastructure. A pro-Iranian hacking group claimed a cyberattack against the Los Angeles County Metropolitan Transportation Authority, alleging access to a rail yard management system. Monitors surfaced the claim within moments of it being posted, giving security teams visibility before the transit authority had confirmed or denied anything. Four days later, the agency still had not issued a public statement.
That four-day silence is not unusual. Official confirmation processes were never built to move as fast as a threat actor who can publish a claim within minutes of an intrusion. A system that flags the claim as soon as it appears, weighs it against available evidence and keeps reassessing as new information arrives turns that waiting period into a head start for response teams instead of a blind spot.
What Autonomous Intelligence Means for Decision Makers
The value of autonomous intelligence extends beyond detecting an incident. The same underlying assessment can help officials understand what the event means for their responsibilities.
An emergency manager may need to know whether disruption to one system could affect public safety or other essential services, how quickly conditions could worsen and whether resources should be staged. A public information officer may need verified information about affected services, what residents should expect and what actions the public should take.
Autonomous agents can continuously translate a common body of intelligence into information relevant to the incident. As the evidence changes, those assessments can change with it.
That reduces the manual research during an incident and gives officials more time to evaluate options, coordinate response and make decisions.
The Governance Question That Matters
Greater autonomy also raises an important governance challenge. If agencies are going to rely on machine-generated assessments during fast-moving incidents, they need clear standards for when to surface assessments and how much confidence to place in them.
Agencies should establish thresholds for corroboration and confidence before deployment. Systems should also distinguish verified observations from analytical conclusions and clearly label projections about what could happen next.
Transparency is essential. Decision makers need to understand why a system reached a conclusion, which evidence supports it and where uncertainty remains.
Human judgment remains essential for consequential decisions. Autonomous systems can take on more of the continuous collection, comparison and synthesis that precedes those decisions, giving people more time to focus on the choices that require experience, authority and accountability.
The agencies best positioned for the next emergency will be those that have already established this operating model, with autonomous systems continuously evaluating emerging information and clear rules governing how machine-generated intelligence earns trust.
In an emergency, the most valuable resource is often time. AI agents can help state and local response teams get more of it.
Tim Miller, Dataminr’s Global Field CTO and Chief Cybersecurity Strategist, has nearly two decades of experience helping public sector organizations adopt and leverage emerging technologies.




