On August 19th 2026, a French Bee A350 approaching Réunion lost satellite navigation after encountering GNSS jamming over the Red Sea; unreliable ground-ranging data then forced a visual approach, according to French investigators. Within three weeks, American trials had put magnetic navigation on an Embraer airliner and Northrop Grumman’s Lumberjack, a one-way attritable aircraft. Read separately, these were aviation and defence experiments. Together, they show how civil and military demand are building the same infrastructure: quantum sensors, maps of Earth’s crustal field and software able to fuse magnetic readings with other data. Its first military use will probably be dependable positioning when satellites fail. A second, narrower spillover could make magnetic anomalies another cue in maritime surveillance, although physics sharply limits what the same method can do against mobile land targets. Governance, therefore, has to reach the maps and software as well as the jammers.

DEVELOPMENT

The pressure is no longer hypothetical. IATA’s 2025 safety report recorded 67% more jamming reports than in 2023 and a 193% rise in spoofing reports. ICAO has separately condemned recurring interference from North Korea and Russian territory as infractions of the Chicago Convention. And the cost does not end at the airport: a British government study estimated that a seven-day national GNSS outage would cost £7.64bn. Such exposure is why alternative positioning has moved beyond a specialist military problem.

Magnetic navigation begins with a fixed feature. Rocks in Earth’s crust produce local variations in the geomagnetic field, which an aircraft can compare with a reference map and then combine with its inertial system. Quantum engineering improves the measurement, not the geology. In September 2026, a Pentagon-backed Honeywell system, after flying an Embraer 170 over the Pacific for four hours and 23 minutes without GPS, reported position accuracy 89% better than a conventional GPS-independent backup. The figure is striking, but incomplete: the announcement did not publish the absolute error, its statistical measure or enough of the test protocol for an independent comparison. Five days later, SandboxAQ announced that AQNav had flown alongside visual navigation on Lumberjack.

The harder constraint may be the map. Existing anomaly datasets, compiled mainly for geology and resource exploration, differ in survey altitude, resolution, age and uncertainty; ocean coverage is particularly thin. America’s Defense Innovation Unit has therefore sought unmanned survey systems to collect new data. Australia awarded A$7.2m to three firms developing navigation for autonomous underwater vehicles, while Britain has invested more than £28m in military quantum PNT and targets airborne deployment by 2030. China, meanwhile, has both a scientific observation base and a military research interest: CSES-02 carries vector and scalar magnetometers, and a PLA-affiliated 2026 study called for ultra-high-resolution maps fused with inertial navigation for unmanned platforms. Uneven though these programmes are, they point towards competition over a complete sensor-map-software stack.

SECURITY IMPLICATIONS

Navigation comes first. A drone that retains trusted coordinates, even after satellite signals disappear, can georegister radar, imagery, signals-intelligence and acoustic observations before passing a coherent track into a command system. The Lumberjack trial, tellingly, placed this capability on an autonomous strike platform rather than a laboratory aircraft.

Such a stack could reduce the defensive value of GNSS denial. Jamming appeals to weaker forces because it is cheap, reversible and can disrupt precision weapons without destroying satellites. A platform able to move among inertial, magnetic, visual, terrestrial-radio and satellite references will be harder to disorient. More subtly, alternative PNT may redistribute alliance dependence. It reduces reliance on American GPS, perhaps, but creates new reliance on whoever owns the regional map, the platform-noise library and the calibration updates. States able to supply those inputs could grant access selectively; buyers denied an update might discover that possessing the sensor is not enough.

The intelligence spillover is more constrained. With a baseline map and the platform’s own magnetic noise removed, the residual field could become another probabilistic feature in a decision-support system. A nearly $1m US Navy programme, for example, is combining commercial magnetometers with an open-architecture classifier for anti-submarine rotorcraft. Yet magnetic detection remains a close-range method: a metallic object’s field falls roughly with the inverse cube of distance. Submarines, mines and large ferrous objects in a bounded search area are plausible targets; mobile land missiles, surrounded by clutter and spread across a wide area, are not equivalent. Magnetic data might confirm an object first identified by radar or imagery. It will not make oceans transparent, and it probably will not find a transporter-erector-launcher by itself.

Here, the distinction among three information layers becomes important: the reference map used for navigation, the compensation data used to remove a platform’s own magnetic signature, and the residual anomaly used to classify an object. Fusion can magnify errors across all three. A stale map or correlated sensor failure, once converted into an automated recommendation, may produce confidence faster than it produces truth. This is the problem governance must actually solve.                                              

GOVERNING THE MAP

The civil layer should remain a common resource. ICAO, through its complementary-PNT programme, should require certified systems to disclose map provenance, machine-readable uncertainty, version history and what happens when one sensor corrupts another. The International Association of Geomagnetism and Aeronomy, which already maintains the International Geomagnetic Reference Field and World Digital Magnetic Anomaly Map, is the sensible technical body for common survey metadata.

Apart from the international cooperation and coordination at International forums, National coordination matters as well. Pakistan’s aviation authority recorded 165 interference events in the Lahore flight-information region in February 2025; after civil, military and telecommunications bodies coordinated, the monthly number fell below 40 by April. The sequence does not prove causation; nevertheless, it shows an important aspect that what reporting and institutional contact can achieve in the uncertain times before new equipment arrives.

In the emerging dynamic in which geomagnetic field is turning into a strategic infrastructure, the security layer needs a narrower rule. First step government can do is to ensure that a map that is derived from geomagnetic field and which is accurate enough for certified civilian navigation stay open. The second step in this regard is that if a government wants to impose certain restriction then it can adopt a targeted approach in which restriction can be applied to, for instance, on raw surveys that cross published thresholds for spatial resolution, collection altitude and the ability to discriminate military anomalies beyond the public safety product. Third and important step government can take is to make reviews (independent assessment of government to classify or withhold data) should be time-limited and independent. Other than aforementioned steps, states should release a degraded civil derivative whenever the raw data are withheld. If these steps are not taken then the fear is that sensitive will become a description of geography itself, and as a result of it, the map enclaves the policy seeks to avoid will form by design.

 Military integration, which lies beyond ordinary civil certification, needs a separate record of accountability. Through the Convention on Certain Conventional Weapons process, states should press for alternative-PNT systems used in targeting to preserve an audit trail showing the map version consulted, its stated error, the other sensors fused with it, any disagreement among those sensors and the point at which a human authorized continued action. Procurement trials should include shifted map tiles, outdated magnetic models, spoofed radio signals, onboard electromagnetic contamination and cases in which several weak cues agree on the wrong location. Indeed, nominal accuracy reveals little about a decision-support system at the moment when its inputs begin to fail together. A weapon that cannot expose the origin of its positional confidence should not be allowed to convert that confidence into a firing solution.

THE CHOICE AHEAD

Earth’s magnetic field is shared. The operational knowledge extracted from it, assembled from survey platforms, correction models and proprietary fusion software, probably will not be. Governments must therefore preserve a continuous and certifiable civil map, confine restrictions to the additional data that pass a published security test, and require weapon systems to reveal how those data shaped a targeting decision. Governance must prevent confidence in a navigation layer from silently becoming confidence in a target.

Author: Nimra Javed, Research Officer, Center for International Strategic Studies, AJK.