Last week someone knocked out 5% of world oil production with a small swarm of drones and cruise missiles, and in doing so, inaugurated “a change in the nature of warfare globally,” to quote The Independent. These were relatively crude drones, too. Let’s pause a moment to imagine what happens if and when sophisticated autonomous drones become cheap enough for even small groups of technically capable insurgents and terrorists to use at scale.
There’s controversy over where and whom the Abqaiq–Khurais attack came from. Even in cases like this, where video exists and wreckage is indicative — “serial numbers on some of the missiles used by the Yemeni rebels in past attacks reveal their Iranian origin” — attribution is hard. What happens if and when autonomous attack drones can be built relatively easily from off-the-shelf parts?
We’re already in the midst of a new arms race. Here’s some video of Indra’s anti-drone system. Here’s Raytheon’s Windshear. Here’s Boeing’s Compact Laser Weapon System. Startups are in on the action too: Dedrone and especially Fortem.
The need for these defenses is obvious. Remember when small, unarmed commercial drones basically shut down the second busiest airport in the UK for days last year?
But, looking forward, will those detect small autonomous drones which hug the ground while avoiding obstacles like a Skydio? Or kamikaze drones which can conceivably defend themselves? Iterations will continue, on both sides, in a classic arms race. One side builds better defenses; the other side builds bigger drones that fly faster/farther and carry more explosive and nosedive onto their targets, or smaller nimbler drones that outswarm defenses; then the defenders upgrade; then the attackers innovate. All in a highly irregular, punctuated way, over the space of years.
That future already seems all but guaranteed. But the bigger question is: even if you can protect hard targets — oil infrastructure, airports, the White House, etc. — how do you defend against the innumerable soft targets out there? What happens when autonomous drones can recognize and target a particular license plate on the highway, and are all but impossible to track back to the attacker?
I’ve been asking these questions for more than a decade now and I still don’t have any good answers. What I do know, though, is that we’d best start analyzing and answering these questions before we are thrown into collective irrational panic and fury by some kind of widespread coordinated drone attack, high-profile assassination, and/or soft-target drone massacre … because if we wait until that hits, we’re pretty much guaranteed to get our answers wrong.