For more than two years, an artificial intelligence system from a Denmark-based company has quietly listened in on Seattle residents calling 911 for medical help. Callers had no idea. The public received no notice. And city officials conducted no formal review before deployment.
The Seattle Fire Department began using live prompts from Corti’s technology in December 2023. The AI analyzes what patients say in real time. It then suggests to dispatchers whether certain calls should skip an immediate ambulance and go instead to a nurse-staffed call center in Texas. Pop-up alerts nudge the humans making the decisions. Dispatchers retain final say. Yet the system operates in the background on every medical 911 call.
This arrangement came to light only after months of questions from reporters. The department confirmed the details in late May. By then, the program had run without transparency for well over two years. Officials still cannot detail exactly how they measure its success. They have not released full contract terms. And they offer no public data on patient outcomes.
The Seattle Times first reported the program’s scope on June 14, 2026. The story revealed that Corti’s tool has contributed to more referrals to the nurse triage line. Department medical director Dr. Michael Sayre once described a 50 percent increase in such referrals. Fire officials later clarified the actual rise reached 32 percent. They insist the prompts simply encourage more consistent application of existing protocols.
But consistency alone does not address deeper worries. The nurse line, managed by the city’s ambulance contractor American Medical Response, operates under relaxed response standards. Ambulances need not arrive quickly. The city does not track wait times for these transferred callers. Earlier Times reporting exposed cases in which patients waited hours. One retiree, Pamela Hogan, called in 2022, endured a delay exceeding 10 hours, and was later found dead in her apartment. Her estate has filed suit.
So the AI now feeds into a system already under scrutiny. Ryan Calo, University of Washington law professor and co-director of the Tech Policy Lab, voiced clear discomfort. “The potential that this company could be a part of the experience that Seattle 911 callers have and they don’t know it, that raises serious concerns,” he said. “I’m troubled on a number of levels.” He added that anyone erroneously routed outside the core 911 system “has a right to know how it happened.”
Experts in privacy and health policy echo those sentiments. Jevan Hutson, who directs the university’s Technology Law and Public Policy Clinic, supports exploring AI for 911 work but insists governance must match the stakes. “I don’t think I’m categorically against the idea of using AI for 911 calls,” he told the Times. “But the governance needs to match the stakes here, and the governance is not matching the stakes.” Franziska Roesner, a computer science professor who co-directs the Security and Privacy Research Lab, put it plainly. “It’s not clear this is a bad thing, but it’s also not clear that it’s a good thing. I think we need to know a lot more.”
The department pushes back on the criticism. Assistant Chief Chris Lombard, who oversees resource management, stressed human control. “The dispatcher still has the ultimate authority to decide how to handle each 911 call.” A Corti spokesperson reinforced the point. “Corti’s role is to support that work, not replace or override clinical judgment.” The company says it does not own Seattle’s data and cannot use calls from the city to train models for other clients. Patient information stays protected, officials maintain.
Yet the relationship between Seattle and Corti stretches back years. It began before the pandemic with tests aimed at detecting cardiac arrest from voice patterns faster than humans could. Early research from Denmark suggested the AI could spot such emergencies more quickly, though it produced more false positives. A 2019 data-sharing agreement allowed the company to train on Seattle calls. Initially the tool served quality assurance on completed calls, reviewing far more interactions than human staff could manage.
Live prompting for triage decisions arrived later. Lombard noted the department rebuilt its call-taking software with Corti’s help around that earlier period. The company has promoted its Seattle partnership aggressively while expanding across the United States. In a 2024 news release, Corti highlighted record growth and quoted Sayre praising real-time prompting for boosting nurse line referrals. The department hosted international visitors for demonstrations last year. Its own annual report described the collaboration in positive terms. A promotional video featured the department’s lead dispatcher calling Corti more than just a solution.
Still, no public metrics exist on whether the AI improves outcomes. No bias audits have been shared. Research on similar systems has shown risks of unequal performance across race, gender, or socioeconomic lines. Seattle’s surveillance ordinance requires review for technologies that observe or analyze people in ways that could raise social justice issues. This program bypassed that process.
Mayor Katie Wilson released an AI vision document in early May. It calls for careful attention to workforce effects, privacy, and bias. A spokesperson for her office told GeekWire the city is developing a public-facing governance framework. That framework will examine whether current uses center human flourishing and the public good. The administration has yet to act on broader 911 and ambulance concerns raised by prior reporting.
GeekWire covered the Times investigation on June 15, 2026, noting the practice fits a wider pattern across Washington. Snohomish County has rolled out AI assistants for both non-emergency and live emergency calls. Kitsap County launched a dedicated AI-powered non-emergency line. A Seattle startup called Aurelian created the tools used there, named AVA and CORA. They answer routine calls instantly or assist dispatchers silently in chaotic situations. Officials in those counties emphasize that humans still handle every true 911 emergency.
Other regions experiment too. Utah agencies test chatbots to filter harassing non-emergency calls. The trend points to growing pressure on 911 centers. Call volumes climb. Dispatchers face overload. AI promises efficiency. But efficiency without oversight carries risks. Cheryl Kauffman, who runs Seattle Patient Advocates, doubts most people would guess AI plays any role in their emergency calls. “I can’t imagine ordinary people in Seattle guessing that AI is involved in their 911 calls,” she said.
Emily Brice of Northwest Health Law Advocates highlighted another angle. Humans must remain at the end of every decision chain. Incentives that push dispatchers to follow AI suggestions too readily could erode that safeguard. The department says no penalties apply for ignoring prompts. That design choice matters.
Questions linger. Has the AI diverted callers who truly needed rapid transport? Do outcomes differ for those routed to the nurse line versus those who receive ambulances? The city tracks neither in detail. Officials say the prompts align with protocols. Yet protocols themselves have drawn fire for allowing long delays.
The program reflects a larger shift. Emergency services nationwide turn to artificial intelligence for triage, quality checks, and call routing. Early adopters like Seattle and Boston use Corti for review after the fact. Live intervention marks a bolder step. Corti positions its platform as support for healthcare providers. Its CEO has written that such systems deliver data-driven help in cities including Seattle since 2019.
But data-driven does not automatically mean transparent. Or accountable. Or equitable. University experts call for public review, clear metrics, and explicit notification. They argue the stakes in emergency response demand higher standards than those applied to routine customer service chatbots.
Seattle Fire Department leaders view their approach as conservative. They point to human oversight and protocol adherence. They note the technology’s role in refining questions asked during calls and improving training at scale. Yet the absence of public metrics leaves residents to trust without verification.
That trust gap widened with the Times investigation. It arrives amid ongoing lawsuits and promises of further review from city leaders. Mayor Wilson and Councilmember Bob Kettle said earlier they would examine 911 and ambulance issues. AI’s hidden role adds complexity.
So far, no changes have been announced. The AI continues to listen. Dispatchers continue to receive suggestions. Callers remain unaware. And the debate over how much technology should shape life-or-death decisions in the dark has only begun.
WebProNews is an iEntry Publication