Public Safety & Emergency ServicesIssue 545 records
Emergency Services
NACo's new 911 workforce brief reports Lyon County, Kansas, now has an AI assistant handling about 35% of calls, with emergencies kept human and staff monitoring every AI interaction, paid for from an unfilled position. A critical review of emergency and prehospital AI urges cautious optimism, noting a dispatch alert that beat dispatchers alone did not improve live recognition. Barrie police recruit dispatchers with an AI mock-call simulator, Ribbit and Canada's NRC flew autonomous SAR aircraft under one commander, and 62 ESCAP members backed satellite-and-AI disaster data sharing.
- Evidence records
- 5
- Cross-source patterns
- 2
- Evidence classes
- 3 independent reporting1 public-sector association guidance1 academic research
- Outcomes
- 3 emerging1 effective1 mixed
- Source freshness
- 5 new this fortnight
- Research completed
- 2026-10-31
Choose a role to see its takeaway beside every record in the ledger.
Patterns across the evidence
Synthesis · Lighthouse Advisory interpretation
Emergency AI is being framed as bounded help with humans watching, not replacement
Lyon County keeps emergency calls human-handled and has staff monitor every AI interaction; the critical review argues the defensible role is governed augmentation and cites a trial where an accurate alert did not improve dispatchers' live recognition; Ribbit's autonomous SAR aircraft still flew with safety pilots under one human commander. Oversight is the common design, but none of the sources measures its workload or how often humans corrected the AI.
Operating questionFor each AI tool you consider, what exactly stays human, who monitors the AI, and how is that time funded?
Supporting evidenceNational Association of Counties (NACo); Lyon County Emergency Communications Center (Kansas)Ramalhete, Oliveira and Quintas; affiliations include NOVA Medical School and the Blood and Transplantation Center of Lisbon (Portugal)Ribbit; National Research Council of Canada (NRC) Drone Innovation Hub
911 centres are applying AI to the staffing shortage from both ends: recruiting and offloading routine calls
Barrie police use an AI-driven mock 911 call to attract dispatcher applicants, while Lyon County uses an AI assistant on routine calls and reports freed time to train two new telecommunicators, funded from an unfilled position. Both target the same workforce gap; neither reports applicant, retention or error data, and the funding choice shows AI can quietly substitute for a hire.
Operating questionIs your staffing plan using AI to attract communicators, to shed routine work, or to avoid hiring, and how would you tell?
Supporting evidenceNational Association of Counties (NACo); Lyon County Emergency Communications Center (Kansas)Barrie Police Service (Ontario) communications centre
Evidence ledger
Full record · every source keeps its link and limitations
NACo 911 workforce brief: Lyon County, Kan., AI assistant 'Betty' now handles about 35% of calls after a four-month pilot, with staff monitoring every AI interaction
NACo's two-part 911 and behavioral health series (Oct. 28) profiles county call-handling and telecommunicator workforce practices. It states PSAPs handle more than 240 million 911 calls a year, including an estimated 24 million behavioral health calls, staffed by more than 100,000 telecommunicators, and that 68% report daily pre-shift and persistent stress. In its Lyon County, Kan., case, the emergency communications center introduced an AI assistant, 'Betty', in December 2025 for routine non-emergency calls (about 80% of volume) amid staffing shortages. After a four-month pilot it manages roughly 35% of calls; emergency calls stay human-handled and staff monitor all AI interactions. The center reports lower workload pressure, better morale, time to train two new telecommunicators, and commissioners redirected an unfilled position's funds to pay for Betty.
Why it matters, evidence and limitations
- Why it matters
- A national county association now presents a non-emergency AI call assistant as a workforce strategy, giving county leaders a peer case with a stated share of calls and a staffing-budget trade-off.
- Evidence and measured results
- Association case summary based on county-reported outcomes. The 35% figure has no stated denominator (all calls or non-emergency calls), period or error data; morale and workload improvements are self-reported without measurement. Vendor and cost are not given.
- Limitations and uncertainty
- Brief case vignette within a broader behavioral health series; figures reported by the center via NACo, not independently evaluated; 'roughly 35% of calls' is ambiguous about denominator; no misroute or caller-outcome data.
Critical review: AI in emergency and prehospital care earns 'cautious optimism', with dispatch AI that beat dispatchers alone failing to improve live recognition in a trial
A critical narrative review (searches of PubMed, Scopus and Web of Science, Jan. 2018 to Sept. 15, 2026) concludes the most credible near-term AI uses in emergency and prehospital care are bounded decision support: triage, stroke imaging, operational forecasting, dispatch support, destination decisions and supervised documentation. It finds the field constrained by retrospective designs, heterogeneous endpoints, limited external validation and a gap between model performance and usefulness. On dispatch, it notes a randomized trial where a machine-learning cardiac arrest alert recognized more arrests than dispatchers alone but did not significantly improve dispatcher recognition in live calls. It recommends local validation, real override, named owners, logging and post-deployment monitoring, and says current evidence supports cautious optimism.
Why it matters, evidence and limitations
- Why it matters
- Gives EMS and 911 buyers an up-to-date, sceptical synthesis to set evidence expectations for dispatch, triage and destination AI, including the lesson that algorithm accuracy does not equal better human performance.
- Evidence and measured results
- Narrative (not systematic) review with a structured search; no pooled estimates; maturity ratings are the authors' qualitative synthesis. Inspected via WebFetch of the full article page because curl returned 403; tables 1-2 were not rendered in the fetched text.
- Limitations and uncertainty
- Narrative review in a new journal; selection of evidence is interpretive; most cited evidence predates this edition; review is about clinical and dispatch AI broadly, with limited fire or disaster coverage.
Barrie Police Service uses an AI-driven 911 dispatcher simulator to recruit communicators, saying it does not replace the assessment process
Barrie police launched a recruitment drive using an AI-driven 911 dispatcher simulator that puts prospective applicants through mock scenarios showing communication, decision-making and high-pressure demands. One resident found it through an Instagram ad, played a scenario with human voices on a mock call and then applied. The communications administrator says it also gives recruiters a sample of candidates' abilities. The deputy chief says it does not replace the broader recruitment and assessment process, and that only a name and email are collected; sharing the simulated call with recruiters is optional.
Why it matters, evidence and limitations
- Why it matters
- A practical, low-risk AI use on the 911 staffing problem: realistic job previews to widen the applicant pool, with minimal data collection.
- Evidence and measured results
- Local news feature built on one applicant's experience and police quotes. No figures on applicants, completion, conversion or retention; vendor and how AI is used in the simulator are not described.
- Limitations and uncertainty
- Single anecdote; no outcome data; it is a police-run centre in Canada, so transferability to U.S. county or fire-based PSAPs is uncertain; it is unclear what part of the simulator is AI-driven.
Ribbit and Canada's NRC fly an autonomous airplane and helicopter under one human commander in a search-and-rescue exercise near Ottawa
Ribbit and the NRC conducted a search-and-rescue exercise near Ottawa in which Ribbit's autonomous Cessna 150 searched for a missing person and adjusted its flight plan using a shared operating picture, then an NRC-operated autonomous Bell 412EP helicopter carried out the evacuation while the airplane monitored overhead. Simulated drones added aircraft to the mission. A single human commander directed real and simulated aircraft; both aircraft flew autonomously with safety pilots aboard. Ribbit says the work followed more than three years of cooperation with the NRC, including contracted testing for civilian and military uses in public safety and other operations.
Why it matters, evidence and limitations
- Why it matters
- Shows multi-aircraft autonomy for search and rescue moving from single drones to crewed-size aircraft coordinated by one commander, a model relevant to state and provincial SAR and wildfire aviation.
- Evidence and measured results
- Defense trade reporting based on company and NRC information; no performance metrics (search time, detection, safety interventions) are given; exercise scale and conditions are not described.
- Limitations and uncertainty
- Single demonstration; company-sourced details; dual civil-military framing; no outcome data; Canadian regulatory context.
62 ESCAP members adopt Bangkok Declaration on space technology, backing satellite and AI integration and data sharing for disaster response through 2030
Khaosod English reports that 53 member states and nine associate members of ESCAP endorsed the Bangkok Declaration on the Use of Space for Sustainable Development at the fifth ministerial conference in Bangkok, guiding the 2026-2030 phase of the Asia-Pacific Plan of Action on Space Applications. The declaration promotes integrating AI, big data, cloud, quantum and satellite communications into space activities, and supports voluntary data sharing under national laws with common data standards, expected to strengthen disaster response and climate monitoring. Thailand's deputy prime minister said combined optical and radar satellites and AI had improved monitoring of floods, droughts and transboundary haze, but the key challenge is turning data into decisions that help people during emergencies.
Why it matters, evidence and limitations
- Why it matters
- International policy signal: governments are committing to AI-plus-satellite early warning and cross-border data standards, while admitting the bottleneck is turning data into emergency decisions; transferability to U.S. SLED is limited to data-sharing and early-warning lessons.
- Evidence and measured results
- News report of a ministerial declaration and a minister's speech; the declaration text itself was not opened; the claim that AI improved monitoring is the minister's, without data.
- Limitations and uncertainty
- Secondary report; declaration not inspected; no AI performance data; ministerial framing; non-binding.
How to read this edition
Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.
- Independent reporting
- Independent reporting with attributable sources but without a formal evaluation design.
- Public-sector association guidance
- Practitioner guidance or an association-supplied case; not independent outcome evidence.
- Academic research
- Research produced through an academic institution or peer-reviewed venue.