DHS Flew “Blue” Drones Against DJI in Manhattan. Three Could Not Hold Position, and the DJI Scores Were Left Out.

The US Department of Homeland Security spent a year flying “Blue UAS” drones, the American-approved alternatives to DJI, through Miami and Manhattan with 26 police and fire pilots at the controls. Its public report, released in September 2026, is blunt: in the final round at Hudson Yards three of the five Blue aircraft could not hold position, one flew away, and one took up to 15 minutes to pair with its controller. The best of them scored 3.7 out of 5. DJI aircraft were flown in the same test. Their scores do not appear in the 47-page document.

Short version: the test confirms what US public-safety agencies have been saying since the FCC Covered List cut off new DJI authorizations on 22 December 2025. The Blue list certifies where a drone was made and how its data is handled; it does not certify that the drone can do the job. The report itself says so. What it does not do is tell American buyers how the DJI aircraft they can no longer buy new actually performed, and that is the part worth reading closely.

What DHS tested, where, and with whom

The work was done by the National Urban Security Technology Laboratory (NUSTL), part of the DHS Science and Technology Directorate, under its SAVER programme (System Assessment and Validation for Emergency Responders). SAVER exists to give first responders hands-on comparisons of equipment they are considering buying. The report is SAVER-T-R-39, Blue UAS for First Responders Comparative Assessment Report, published in September 2026.

Testing ran from June 2024 to July 2025 in three rounds, the second in Miami in November 2024 and the third at Hudson Yards in Manhattan in July 2025, the kind of glass-and-steel canyon where GPS reflects, radio links fail and a drone’s own sensors decide whether it stays put. The evaluators were 26 public-safety pilots from 13 states, including the NYPD, FDNY, Boston Fire, the Los Angeles Police Department, Michigan State Police, Texas DPS, the Coast Guard and Customs and Border Protection.

Each aircraft was scored from 1 (met none of the evaluators’ expectations) to 5 (exceeded all of them) on nine criteria: visual-light camera acuity, thermal camera acuity, flight duration, command-and-control link quality, latency, time to redeploy, deployability, ease of use and the ground-control interface.

What happened to the five Blue UAS

Three sentences from the report explain the table better than any commentary. On navigation: “GPS cannot be relied on in dense cities.” On the radio link: “the link between controller and aircraft is inconsistent.” On what the Blue stamp actually means: “a Blue designation says nothing about whether an aircraft can do the job.” A fourth line, quoted by the Drone Service Providers Alliance, puts it in procurement language: “A cybersecurity or procurement designation is not a substitute for operational capability.”

None of this is an argument that the Blue aircraft are bad drones. The Skydio X10D is the same airframe that Skydio built its 2026 product line around, and a 3.7 in a Manhattan canyon is a respectable score. The point the evaluators keep returning to is reliability of the basics: holding position without GPS, keeping the link up, launching in under a minute. Those are the things a fire chief notices.

The numbers that were left out

The public report confirms that “covered nation platforms were flown alongside the Blue UAS at Hudson Yards” and says their results are available to public-safety agencies on request by email. They are not printed. A separate addendum, SAVER-T-R-40 (July 2026), is marked For Official Use Only.

Single source, not independently verified by us: on 1 October 2026 DroneXL reported that it had obtained the addendum and published the withheld figures. According to that report, the DJI Matrice 4T scored 4.6 overall at Hudson Yards, with 4.9 for visual-light acuity, thermal acuity and link quality, and the DJI Matrice 30T scored 4.3, with the best time-to-redeploy score of the field at 4.7. The same account records a Matrice 30T losing both satellite and vision positioning at night in a narrow street and drifting into a structure. DroneXL has not republished the document itself, and DHS has not confirmed the figures publicly. We treat them as reported, not established.

DHS’s public response did not dispute the numbers. It restated the policy: NUSTL “unequivocally does not endorse equipment on the FCC Covered List” because of “unacceptable risk to U.S. national security,” and the SAVER assessment examined operator experience only, not security.

Score against price

Read across a row and the procurement problem is plain. The highest-scoring aircraft in the test cost less than half of the highest-scoring Blue aircraft, and a third of the aircraft at the top of the price range could not be scored at all in the final round. An agency that buys Blue is paying more for less measured capability, and it is doing so for a reason the test was explicitly not designed to weigh: security.

This was not a security test, and nobody says it was

It would be easy to read the report as a verdict on the Covered List. It is not, and DHS is right to say so. SAVER measured what a pilot sees: image quality, link stability, how long it takes to get airborne. It did not inspect firmware, data flows or supply chains, which is what the FCC’s December 2025 decision and the earlier Commerce and Defense actions were about. A drone can be excellent to fly and still be one a government does not want on its networks.

What the report does is remove one argument from the debate. Before it, the case for the Blue list could be made on both security and capability. After it, the capability half is gone, in writing, from a DHS laboratory. That is why the Drone Service Providers Alliance, a commercial-operator group that normally stays out of public-safety procurement, asked its members on 27 September to write to Congress with a single request: name a federal agency to test the security of every drone flown in US airspace, and let those results, not the country of manufacture, decide what agencies may buy. The Section 1709 review that was supposed to do that work missed its 23 December 2025 deadline, which is how the Covered List listing happened by default.

What it means if you are buying

• US public-safety agencies. Nothing in the report changes what you are allowed to buy. Matrice 4T and 30T units authorized before the cut-off are still legal to purchase and fly, but state laws (Florida, Arkansas, Tennessee and others) and federal grant conditions often rule them out regardless. The US drone ban checker shows the federal status model by model; check your state’s list separately.
• Anyone weighing a Blue aircraft. Ask the vendor for the SAVER scores by criterion, not the overall number. The failures at Hudson Yards were specific: position hold without GPS, link stability at short range, pairing time. Those can be tested in an afternoon in your own downtown before a purchase order is signed. For the two aircraft at the centre of the report, our Skydio X10 vs DJI Matrice 4T comparison sets the test scores against the published specifications and prices.
• Buyers outside the US. The Covered List does not apply to you, and the test is simply a data point in favour of the Matrice 4T. Our Matrice 4T vs 4E comparison covers the thermal-or-not decision; the Matrice 400 is the step up when payload matters more than size.
• Everyone watching 13 October. The FCC’s component rule, which extends the Covered List to foreign-made parts in otherwise domestic drones, takes effect on 13 October 2026. It makes the Blue list more important, not less, which makes the capability gap this report documents more expensive to ignore. Details in our component rule explainer and the US drone ban hub.

What we still do not know

• The Hudson Yards scores for the Teal 2, Freefly Astro and Ascent Spirit. The public report describes their failures but does not score them for that round.
• Whether the addendum used exactly the same nine criteria and the same evaluators for the DJI aircraft. DroneXL’s account implies it did; the document is not public.
• Whether Skydio’s single-battery shipments, noted in the report for October 2024, have been resolved for current X10 orders.
• How the withheld results are being used. DHS says agencies can request them, which means the people making purchasing decisions can see numbers the public cannot.

We will update this page if DHS publishes the addendum, if any manufacturer responds on the record, or if Congress acts on the DSPA request. Last verified: 5 October 2026.

Frequently asked questions

What is the Blue UAS list?

A list of small drones vetted by the US Department of Defense (through the Defense Innovation Unit) as compliant with federal supply-chain and cybersecurity rules. It covers where an aircraft and its components are made and how its data is handled. It is not a performance certification, a point the DHS report makes explicitly.

Which drones did DHS test?

Five Blue UAS: the Skydio X10D, Parrot ANAFI USA GOV, Teal 2, Freefly Astro and Ascent AeroSystems Spirit. DJI Matrice 4T and Matrice 30T aircraft were flown alongside them at Hudson Yards; their scores were withheld from the public report.

How did the Skydio X10 do in the DHS test?

It was the best-performing Blue aircraft, scoring 4.0 in Miami and 3.7 at Hudson Yards. Evaluators noted some video pixelation and link losses at short range.

What did the DJI Matrice 4T score?

According to the addendum obtained by DroneXL, 4.6 overall, with 4.9 for both camera criteria and for link quality. DHS has not published or confirmed that figure; we report it as a single-source number.

Does the report mean the DJI ban will be reversed?

No. The report measured flying performance, not security, and DHS repeated that it does not endorse Covered List equipment. Reversing or narrowing the restriction would take action by the FCC or Congress. The DSPA is asking Congress for a uniform security test of all drones; nothing has been scheduled.

Can a US police or fire department still buy a Matrice 4T?

At the federal level, yes: units authorized before 22 December 2025 remain legal to buy and fly, and dealers still hold stock. In practice many agencies cannot, because state laws or federal grant conditions prohibit Covered List equipment. Check your state’s rules before relying on the federal position.