China manufactures a new drone every second. Let that sink in. Every second, another low-cost, high-agility platform rolls off a production line, ready for surveillance, disruption, or kinetic attack. These autonomous unmanned aerial systems (UAS) are becoming harder to detect, harder to predict, and exponentially harder to neutralize using conventional means.
In military theaters and civilian airspaces alike, these silent operators are increasingly described as “ghosts”—emerging from the electromagnetic fog, undetectable until it’s too late. But in this new era of spectral warfare, artificial intelligence is offering us a powerful way forward.
The Software-Defined Battlefield
Our legacy systems were built to track large, human-piloted aircraft using radar and RF patterns, but they’re essentially blind to small, autonomous drones. The electromagnetic spectrum hasn’t seen real software innovation since the Cold War. Hardware has improved, sure, but the software side is only now catching up, forced by what we’re seeing in Ukraine and the Middle East.
AI changes everything. Instead of waiting for known signal types, AI watches how things behave—pulling together data from radar, acoustic, visual, and RF sensors all at once. And we’re shrinking these capabilities down. The future of counter-UAS is becoming both smarter and smaller.
At Zighra, we’ve focused on the intersection of sensors, behavioral anomaly detection, and explainable AI—bridging digital AI with physical AI. Our journey into this space began through work with the Coast Guard on jamming and spoofing challenges, where we discovered that the intersection between digital intelligence and physical sensors could detect AI-powered attacks in under three seconds. Our systems create adaptive behavioral models that flag anomalies as they emerge—essential when China is manufacturing a drone every single second.
Autonomy and Adaptive Threats
Autonomous drones don’t need GPS or remote control. They fly pre-programmed routes, dodge jamming, and navigate without any external help. Traditional radar can’t see them, and there’s nothing for RF detection to pick up. In Ukraine, AI-enabled drones are achieving 70-80% strike accuracy—three to four times higher than manually piloted systems¹ – demonstrating the capability leap that AI brings to modern aerial warfare.
The speed of innovation here is staggering. I’ve watched this pattern repeatedly: defense systems solve one problem, adversaries immediately pivot to something new. When you’re producing a drone every second, you can afford to constantly experiment—new signatures, new parameters, new attack vectors. It reminds me of fraud detection in the financial world, where we’ve been for years. Fraudsters adapt constantly, and when generative AI arrived, that adaptation went into overdrive.
AI-enabled counter-UAS systems address this through behavioral inference. By fusing data from passive sensors, these systems begin to “see” patterns that machines without AI would ignore. Like tracking shadows instead of the object itself, AI allows us to detect intent – something no static radar dish or spectrum scanner can do alone.
Beyond Detection: Intelligent Response
Detection is only half the battle. The other half is response—and this is where AI truly shines. Modern counter-UAS strategies are moving toward precision neutralization. Think of systems that can deploy directional jammers only when a confirmed threat is within a critical zone, or AI-enabled effectors that scale their response based on target type and proximity to civilians.
Current defense mechanisms—jamming, spoofing, kinetic destruction, or shooting nets—are becoming insufficient against programmable, AI-powered drones that adapt in real time. Traditional procurement cycles that deliver systems 8-10 years after ordering simply cannot keep pace with threats that evolve weekly.
At Zighra, our approach enables real-time adaptation at the edge. We don’t need to go back, recode, reprogram, and push updates—that kill chain is too slow. Our systems continuously adapt to evolving threats through behavioral modeling, detecting new attack patterns on the fly.
The Arctic Challenge
This challenge hits home for Canada, especially in the Arctic—a region warming four times faster than the global average and increasingly contested by global powers. You can’t just grab an off-the-shelf counter-drone system and deploy it up there—the environment is too extreme, the distances too vast. Our small population and massive landmass create a unique security puzzle.
We’re investing in over-the-horizon radars and various detection systems, but when small drones come in swarms, flying under the radar, these big systems miss them entirely. Satellites are looking for large objects. This gap is exactly where adaptive AI systems make sense—creating meshed networks that blanket huge areas while constantly watching for electromagnetic anomalies that traditional systems simply can’t see.
Procurement Revolution
Perhaps the most significant challenge is institutional rather than technological. Defense organizations worldwide are recognizing that their traditional procurement methods are failing against rapidly evolving autonomous threats. Ukraine has learned to procure and adapt extremely quickly under pressure, but established defense departments struggle to transform their entrenched processes.
Consider Canada’s recent $2.49 billion investment in advanced drone systems – with first delivery expected in 2028 and full operational capability not until 2033². While Ukraine procures and deploys thousands of drones within weeks, our traditional procurement timeline spans nearly a decade. This disconnect between innovation speed and acquisition cycles represents a fundamental strategic vulnerability.
Canada has tremendous AI and software talent. We need to leverage this to build our unique defense industrial base capability. As we play a bigger role globally from a geopolitical perspective, we must defend our own Arctic while contributing meaningfully to allied defense.
Ethics and Explainability
What happens when an AI system gets it wrong and targets a medical delivery drone instead of a threat? They can fly identical paths, and the difference between them isn’t always visible.
Explainable AI becomes essential in these scenarios. We’ve made transparency non-negotiable in our defense applications at Zighra. Our systems provide detailed explanations for every decision, quantify their uncertainty levels, and alert operators when confidence drops. Every decision pathway gets logged for review. Autonomous systems without this level of accountability become reckless rather than innovative.
The Path Forward
We must ask: Are we building safer airspace or accelerating an arms race in autonomy? The answer depends on our principles. If we prioritize collaboration, transparency, and shared norms of AI development in defense, the future can be one of safety and resilience.
The drone threat will evolve. So must our defenses. But in guarding against these ghosts, we must remember that the goal goes beyond neutralization. We need to outthink, outmaneuver, and outlearn our adversaries.
With AI on our side—and with the right ethical frameworks – we can not only meet the challenge but shape a future where autonomy protects, rather than threatens, our skies.
Deepak Dutt is the Founder and CEO of Zighra, the AI platform for Cyber & Electronic Defence. A pioneer in AI-driven security, he has led innovations in adaptive threat detection, behavioral analytics, and real-time cyber & electronic defence. With 14+ patents in AI and cybersecurity, Deepak collaborates with defence, government, and critical infrastructure organizations to develop autonomous, mission-ready security solutions that counter AI-powered cyber warfare, electronic interference, and identity-based threats.
