AI Dogfighting? YES!

[Signal Witch]

Northrop Grumman’s “Beacon” Platform: Quiet Testbed for AI Dogfighting?


 
Northrop Grumman recently announced the first phase of its Beacon program, a flight-test initiative centered on autonomy integration. It uses the Scaled Composites Model 437, also known as Vanguard, to trial real-world autonomy stacks from a group of handpicked partners.
The goal is to move autonomy out of labs and simulations and into live airborne environments under mission-like conditions.
The first wave of selected companies includes:

• Shield AI, known for its Hivemind AI pilot
  
• Red 6, specializing in augmented reality combat overlays
  
• SoarTech, Applied Intuition, Autonodyne, and Merlin, all offering various navigation, flight control, and combat logic technologies
  

This is not early-stage DARPA research. These are flight-ready autonomy systems being tested on a flying platform that resembles a scaled-down business jet. According to Northrop, the architecture is modular, allowing quick integration of different autonomy stacks. Live integration flights are planned to begin this year.
 
What stands out here is the mission maturity. Red 6’s involvement points toward adversarial training. Shield AI and Merlin already hold autonomy contracts with the Department of Defense. Beacon may be the proving ground for AI-enabled dogfighting, collaborative swarm tactics, or manned-unmanned teaming, without requiring the exposure of frontline systems like NGAD or CCA platforms.
 
Given Northrop’s heavy classified investment into autonomy over the past several years, Beacon may represent the middle-tier transition point between lab environments and deployment on operational or soon-to-be-fielded airframes.
 
What do you think?

Could Beacon be quietly developing AI dogfighting capability?

[Zaphod58]

I remember watching the AI vs human simulated dogfights several years ago.  IIRC the timeline they released during that had them flying physical unmanned aircraft around now.