Spotlight

NTSB Investigates GPS Jamming in New Mexico Plane Crash

July 2026: A series of aircraft near-misses due to GPS jamming near the U.S. Army-operated White Sands Missile Range may have culminated in the fatal May 2026 crash of a civilian medevac plane and subsequent month-long wildfire. The U.S. National Transportation Board (NTSB) has issued a preliminary report investigating the role White Sands may have played in the accident, and marks the first report in which the NTSB has held GPS jamming as a factor.

The RNL’s Dr. Todd Humphreys, who has participated in GPS testing at White Sands, says that some tests in the region involve spoofing GPS signals. To avoid these signals misleading civilian aircraft, the Army placed powerful GPS jammers along the range perimeter, drowning them out. Dr. Humphreys remarked that it is likely this preventative ‘cover jamming’ that the medevac plane experienced in May, saying that “the jamming coming from White Sands is intense. It goes all the way up to low Earth orbit.”

Furthermore, FAA-certified navigation hardware of the kind used on civilian aircraft is heavily reliant on the L1 GPS frequency. These systems — unlike modern multi-frequency or multi-constellation GNSS receivers — are more susceptible to GPS interference. Dr. Humphreys noted that the “FAA certification process is so onerous and conservative that you’re dealing with aviation-grade receivers that are more than a decade old in technology.”

Read more about the crash, as well as White Sands’ effects on American airspace, with IEEE Spectrum.

The Subtle Importance of GPS

July 2026: In June 2017, Russian civilians and military personnel alike became casualties of a mysterious, meandering cloud of GPS spoofing. It was later revealed that this spoofing came from Russian President Vladimir Putin’s security detail for protection from drone strikes.

The RNL’s Dr. Todd Humphreys, who built the first civilian-operated GPS spoofer, had predicted such GPS incursions could occur. Speaking about his spoofer in a 2012 TED Talk, Dr. Humphreys saw “airplanes and ships veering off course, with the captain learning only too late that something was wrong. I saw the GPS-derived timing of the New York Stock Exchange being manipulated by hackers. You can scarcely imagine the kind of havoc you could cause if you knew what you were doing with a GPS spoofer.”

On Slate’s Decoder Ring podcast, Dr. Humphreys remarked, “Thousands and thousands of papers have been written since, and I think it was all to the good, because when spoofing arrived in the wild… we were much better prepared than if we had just sat around saying, ‘I’ll keep my head in the sand.'” GPS spoofing has grown increasingly prevalent in the wartime theater, its effects spilling into civilian life in Europe and the Middle East.

Listen in with Slate’s Willa Paskin on her episode of Decoder Ring for a history of GPS, its evolving role on the global stage, and an interview with Dr. Todd Humphreys (beginning at 27:22). A transcript is also viewable here.

Machine Vision Arms Race Continues in Ukraine

July 2026: On the Ukranian battlefront, Russian military trucks have taken on a striped camouflage to trick AI models into not recognizing them. This “dazzle camouflage,” once used during WWI to break up naval ship silhouettes against the horizon, is now being used to fool machine vision models. “Because zebra-striped trucks are unlikely to appear in training data, an AI that encounters one in the real world may not realize what it is looking at,” says the RNL’s Dr. Todd Humphreys.

Machine vision models often rely on very specific features to recognize an object, leading to a problem known as “brittleness.” The Russian military aims to obscure these expected features with unexpected ones. Though battlefield drones are still flown by remote human operators, AI will soon take on more of the work as the drones proliferate, making these tactics likely to grow common.

Read more about the arms race pitting increasingly-sophisticated machine vision systems against clever methods to subvert them on The Economist.

RNL GPS Spoofing Retrospective

June 2026: GPS has become seamlessly enmeshed into almost all areas of everyday life. “We’re almost blindly reliant on it,” the RNL’s Dr. Humphreys said. “It’s built deeply into our systems and infrastructure. Some call it ‘the invisible utility.'” In 2007, spoofing GPS was considered too difficult for the layman to do. Renowned GPS expert Logan Scott, however, wrote that mass spoofing was only a matter of time.

A year later, Dr. Humphreys proved the claim by building the first publicly-recognized GPS spoofer. Over the following years, this proof-of-concept grew into several impactful RNL spoofing demonstrations, such as subtly veering the White Rose of Drachs superyacht a mile off-course and fooling an $80,000 drone into plummeting to the ground. The real-world geopolitical implications were clear: “Spoofing a GPS receiver on a [drone] is just another way of hijacking a plane,” Dr. Humphreys argued.

Read more of the RNL’s retrospective on GPS exploitation on The Walrus, which excerpts Katherine Dunn’s new book – Little Blue Dot: How GPS Shaped the Modern World.

Russian Satellites Caught Jamming GPS in Europe

June 2026: A recent paper from the RNL found Russian satellites responsible for seconds-long GPS outages across Europe since 2019. Corroborated by research at Spanish company GMV and confirmed by U.S. Air Force officials, the RNL’s Dr. Todd Humphreys and RNL alumnus Dr. Zach Clements held the Russian EKS constellation responsible.

Though these fleeting bursts of interference haven’t caused major disruptions, they are among the earliest known cases of GPS interference from space. On Earth’s surface, such interference would be limited to a small radius. Malicious jamming from space, however, could escalate its effects to an entire continent.

Regardless of who is responsible, the RNL’s Dr. Todd Humphreys said that these emissions should be “a wake up call for all of us.” For a detailed discussion of the interference, read more here.

RNL Featured on Veritasium


June 2026: Veritasium interviewed the RNL’s Dr. Todd Humphreys about his and RNL alumnus Dr. Zach Clements’ recent work tracking GNSS interference, exploring the journey and techniques behind identifying the powerful source of transient interference affecting continental Europe, Greenland, and Canada since 2019.

Dazzle Camouflage Confuses AI Drones

June 2026: As Ukraine ramped up an AI-assisted drone campaign to strike Russian logistics equipment, the Russian military opted for a unique form of camouflage. Russian military vehicles have been seen coated with vivid stripes in efforts to baffle computer vision models.

Experts confirm that such Zebra patterns are capable of throwing optical hunting systems off the scent. The RNL’s Dr. Todd Humphreys said this “dazzle paint pushes the vehicles ‘out of distribution’ — they no longer look, to the AI classifier, like the images it was trained on.”

“But any specific paint job would have a brief shelf life,” he continued, as “the AI classifier would need to be retrained on thousands of images of vehicles with dazzle paint, after which the paint scheme could be changed [—] and the cycle starts again.” Russian forces are thus continuously testing new camouflage, and Ukrainian forces confirm that they, in turn, are continuously adapting.

Find out more about dazzle camouflage and the Ukrainian response here.

Dr. Clements and Dr. Haydon Successfully Defend!

May 2026: Dr. Zach Clements (GNSS Interference Detection and Geolocation) and Dr. Tucker Haydon (Trusted Inertial Terrain-Aided Navigation (TITAN)) have successfully defended their PhD dissertations. The RNL gathered at Dr. Humphreys’ home to celebrate. Congratulations to Zach and Tucker!

Dr. Zach Clements
Dr. Zach Clements (Right) with Dr. Todd Humphreys (Left)
Dr. Tucker Haydon
Dr. Tucker Haydon

Researchers Can Still Exploit Starlink for PNT, Despite Removal of “Starlink PNT” Service

May 2026: Weeks before the SpaceX IPO, Starlink has shut down a GPS alternative that most customers never knew about. Other customers found access to built-in location features hidden away in a Debug menu to get coarse position estimates — a feature of immense value to those in regions with rising GNSS interference.

Starlink is of key interest to low Earth orbit (LEO) PNT research. “The beauty of Starlink as a backup to GNSS is that it’s such a different system—frequencies 10 times higher, bandwidths 10 to 100 times wider, power 100 to 1,000 times stronger, [and] satellites 100 times more proliferated,” said the RNL’s Dr. Todd Humphreys. Current PNT solution accuracy is still limited compared to GNSS. The RNL demonstrated mock fused LEO GNSS — with Starlink satellites supplying real-time clocks and orbit corrections — with solutions accurate within 10 meters, albeit with minutes-long processing delays. Dr. Humphreys asserted that “we’re now refining our techniques so it can be done in tens of seconds rather than tens of minutes.”

Starlink may have shut down its “cheat code,” but many other methods of its exploitation remain. For further discussion of PNT solutions using Starlink and other LEO satellite constellations—including a feature from RNL alumnus Dr. Zak Kassas—read more with Ars Technica.

Starlink to End Access to ‘Starlink PNT’ Service

April 2026: Starlink users will soon be unable to access their location on their communications terminals, leaving many confused and upset. The gRPC API on Starlink terminals lends users an opportunity to pull the coarse location generated by the terminal, even in the presence of spoofing and jamming. “Starlink PNT, as it’s called, has a cheat code for those who know [it],” says the RNL’s Dr. Todd Humphreys. “And it has been getting increasingly accurate, from 100 meters a few years ago to about 20 meters now.”

As to why Starlink decided to end Starlink PNT, Dr. Humphreys has a few theories. “SpaceX might not want the liability of providing a location service whose accuracy is decent but variable… They also might be preparing to launch a fee-based version of Starlink PNT and so are ending free access.”

Read more with Inside GNSS.

1 2 … 31