The drone gets there first. Then what?

Eight months into a trial that began in Islington, the Metropolitan Police told the BBC in June 2026 that it was running three drone bases across London, with nine drones attending around 200 incidents each week. The Commissioner, Sir Mark Rowley, set out plans to scale that into what he called a "truly integrated, city-wide drone network".

That is a real deployment with countable parts, and it is worth separating from the thing usually written about drones and emergency services, which is a product launch. What follows is what operators have published about their own programmes, what they measured, and what is still contested.

The model, and where it varies

The London arrangement is a piece of architecture rather than a drone. The Met has described drones housed in weather-proof boxes on the roofs of selected force buildings, launched only in response to a 999 call, flown remotely from the force control room, and streaming footage to officers on the ground. The force said drones would arrive on scene within two minutes, and that footage not needed as evidence is deleted after 28 days.

It also made a comparison worth attributing rather than repeating. The force told the BBC that the new drones were "quicker, quieter, cheaper and more environmentally friendly than police helicopters, while delivering the same operational results". That is the Met's claim about the Met's programme. No independent measurement of it has been published.

Not every service works this way. Helsinki City Rescue Department has flown drones for fire monitoring since 2020, and the picture in the European Union Aviation Safety Agency's record of the programme is a modest one: drones with thermal cameras and lights, deployed once firefighters are already at the scene, showing hotspots, the direction the fire is spreading, and where trapped people are. That same record notes that "until now, the drones are flown within visual line of sight". A rooftop dock with remote launch is one model. A thermal camera flown by the crew at the incident is another, and it is the more common one.

What operators have measured

The most careful measurement in this field is Swedish, and it is careful because it reports what did not work.

Since 2020, Karolinska Institutet, Region Västra Götaland, the national emergency operator SOS Alarm and a drone operator have sent defibrillators to suspected cardiac arrests across an area of roughly 200,000 people. In the first study, over four months in 2020, drones took off for 12 of 53 alerts and delivered a defibrillator in 11 of those, arriving before the ambulance in seven, with a median gain of 1 minute 52 seconds. The researchers then recorded the outcome plainly: "No drone-delivered defibrillators were attached to the patients before the ambulance arrived." Arriving first and being used are different achievements.

The follow-up, published in The Lancet Digital Health in 2023, is where the chain gets interesting, and every link in it has its own denominator. Drones delivered a defibrillator in 55 cases of suspected cardiac arrest. In 37 of those, delivery beat the ambulance, by a median of 3 minutes 14 seconds. Of the 18 cases that turned out to be actual cardiac arrests, the caller managed to use the device in six. A shock was advised in two. One patient survived.

That is not a disappointing result. It is what an honest measurement of a young intervention looks like, and it is the only figure in this article that arrives with a published method attached.

At a different scale, Surf Life Saving NSW reported in June 2026 that its shark surveillance programme had flown more than 100,000 flights in a year, alongside a New South Wales government commitment of a further 34 million dollars and year-round coverage of Sydney beaches from 1 July 2026. Its chief executive said the programme had "identified and prevented over 2000 sharks interacting with swimmers and surfers". No method for that second figure appears on the page. The state Premier's framing was more careful: "While no one can ever promise no shark interactions, this investment is about putting more eyes in the sky so we can spot sharks earlier and give people a clear heads-up when they're in the water."

In the United Kingdom, HM Coastguard put its first operational uncrewed flight alongside helicopters, planes, boats and cliff rescue teams from Caernarfon in 2021, after earlier work with the Civil Aviation Authority on flights beyond visual line of sight.

Vendor claims, and testable ones

Alongside the operators sit the companies selling to them, and their numbers travel further than anybody's.

Flock Safety, which announced its Alpha system in October 2025, publishes a set of performance figures for its drone as first responder product: 86 seconds to reach calls for service, one call in five resolved without dispatching patrol, an 89 per cent increase in subjects located. The same page says Alpha "reads license plates from up to 2,000 feet away" and that thermal and low-light optics "remove guesswork".

Read those as the company's claims, because that is what they are. No methodology accompanies them. One gap is worth naming: the company says Alpha "boasts the longest flight time of any DFR quadcopter drone" and publishes no figure for it anywhere. A superlative with no number behind it cannot be checked against a rival product, or against how long a shift lasts.

A related correction, because the opposite is often repeated: this class of system is not confined to public agencies. The company's own pricing page answers the question directly, saying that "in addition to public agencies, we help businesses, HOAs, houses of worship, and organizations of all sizes protect people and property and reduce crime", and it sells a separate automated drone security product aimed at retail centres, logistics yards and healthcare campuses.

What is disputed

Three things, and none of them is settled.

Cost has the hardest numbers, and they come from Chula Vista in California, the first United States department authorised to fly beyond visual line of sight to 911 calls across a whole city. Reporting by inewsource in August 2026 put a proposed five-year agreement at 400,000 dollars in the first year and 700,000 dollars annually after that, with the department saying the change would save an estimated 1.2 million dollars a year. The city later backed away from the proposal.

Oversight is where the argument is sharpest. Big Brother Watch's Jake Hurfurt told the BBC that drones may have a role in providing quicker updates to evolving emergencies, but that the Met "must not use them as constant flying surveillance cameras", and that the force had "point blank refused to disclose to the public any documents outlining how, why or when it uses these drones". Assistant Commissioner Laurence Taylor had put the force's position in 2025: "This isn't surveillance. This is to support policing operations on the ground." Flock's material takes a similar line, stating that its system "only activates in response to specific calls for service". In San Diego, Seth Hall of the group San Diego Privacy described the model to inewsource as "taking all of the problems that we have with that company, taking them off mounted poles, and launching them into the air".

Effectiveness is disputed mostly because of who is counting. Set 86 seconds beside 55 deliveries and one survivor, and the difference is not that one programme works and the other does not. It is that one of those figures has a published denominator and the other does not.

Questions that have answers

Anyone weighing one of these programmes, in a council chamber or at a public meeting, has better questions available than whether the technology is impressive.

Does the service publish its flight logs, and can they be read without a records request? What is the retention period for footage that is never used as evidence, and who audits it? What is the stated trigger for a launch, and is anything other than that trigger permitted? Who authorised any flights beyond visual line of sight, and on what conditions? And when a performance figure is quoted, what is its denominator, and who measured it?

Every one of those has an answer that exists somewhere. Whether a drone can get there first has largely been settled. What happens in the minutes after it arrives, and who is allowed to know, has not.