Evolving Threat
Monitor

Are drones becomingthe perfectthreat?

Let’s keep watching.

Report for

Six of ten traits have been demonstrated across different systems. Race results still have limits, and a drone combining all ten remains unverified.

What does this mean? →
Abstract red and yellow shapes: a circle, an orbital form, a diamond and a soft star.
  1. 1Autonomy
  2. 2No radio link
  3. 3Swarm coordination
  4. 4Adaptation
  5. 5Speed
  6. 6Persistent presence
  7. 7Low visibility
  8. 8Resilience
  9. 9Scale & replenishment
  10. 10Evolving versions

Ten traits

Last reviewed · Google Sheets ↗

Choose a trait to see what is known, what remains uncertain and where the information comes from.

  1. Makes decisions onboard

    Research drones make navigation decisions onboard. MonoRace adds a competition example, but these tasks do not establish unrestricted autonomy.    Shown in practiceConfidence: Medium

  2. Works without radio control

    DARPA and Qamcom say their systems can work without the operator’s radio link. Clear, separate link-loss test details are still missing.    Reported, not yet verifiedConfidence: Low

  3. Coordinates as a swarm

    Small research swarms coordinate their flights. Racing side by side is different: KAIST reports that its race setup lacked dedicated avoidance of other drones.    Shown in practiceConfidence: Medium

  4. Adapts to changing defenses

    We have not verified a test showing a response to changed defenses. Avoiding obstacles is a different capability.    Not enough informationConfidence: Low

  5. Moves fast

    Autonomous racing drones have completed timed races. KAIST also reports a crash during a faster attempt; race times do not establish a defense’s response time.    Shown in practiceConfidence: Medium

  6. Sustained airborne presence

    Zephyr demonstrated a long flight. Reliable repeated missions remain unverified.    Shown in practiceConfidence: Medium

  7. Is hard to detect

    Ornadyne says detection systems classified O1 as a bird. Public reports omit the comparison and test conditions needed to confirm how difficult it is to detect.    Reported, not yet verifiedConfidence: Low

  8. Continues after a failure

    A research drone followed a route with one propeller missing. This demonstrates a limited task after a fault, without independent replication or proof of resilience to other damage.    Shown in practiceConfidence: Medium

  9. Operates at scale and replaces losses

    Production figures and swarm trials do not show how losses of a specific system are replaced.    Not enough informationConfidence: Low

  10. Evolves through new versions

    Three Skydio X10 software releases document regular updates. They do not establish that each update increases the threat.    Shown in release recordsConfidence: Medium