Scarlet Dragon Links Military, Industry to Test Artificial Intelligence for Warfighters

Exercise Scarlet Dragon is the U.S. Army XVIII Airborne Corps’ main innovation and experimentation exercise. Its mission is to bring soldiers together with other services, government agencies and industry to test new technologies—especially AI, data‑sharing and sensor‑to‑shooter networks—to close real battlefield “capability gaps” such as faster targeting and better drone defense. It began in 2020 as a small tabletop drill in the Corps headquarters basement and has grown into a recurring (about three‑times‑per‑year) joint event spanning multiple services and partners, used to refine long‑range fires, data‑centric warfare and now counter‑UAS and drone “mothership” concepts.

Counter‑unmanned aircraft systems (counter‑UAS or C‑UAS) are tools and methods used to protect people and sites from unwanted or hostile drones. They usually combine sensors (like radar, radio‑frequency detectors, cameras or acoustic sensors) to detect and track drones, software—often with AI—to recognize and classify them, and effectors to respond, such as jamming the drone’s control signals or GPS, taking control of it, or in some cases physically shooting it down. The goal is to see small drones early, understand whether they are a threat, and safely stop them before they can cause harm.

The article itself does not list every algorithm, but it shows several concrete AI‑enabled capabilities being tested:

• Maven Smart System (MSS): an AI‑powered software platform from the National Geospatial‑Intelligence Agency that pulls together data from many sensors and databases and uses machine learning to rapidly generate targeting information for systems like HIMARS rocket launchers.
• Integrated air‑defense “shared picture”: AI‑assisted fusion of data from Sentinel radars, SGT STOUT short‑range air‑defense vehicles, Apache helicopters and drones into a single, faster early‑warning picture for troops.
• Drone “mothership” autonomy: an autonomous unmanned aircraft acting as a “mothership” that launches and coordinates multiple smaller drones for wide‑area sensing and data sharing. More broadly across the Scarlet Dragon series, MSS uses AI to ingest satellite and other imagery and automatically detect, classify and prioritize targets, cutting the time and staff needed for long‑range artillery targeting.

The Scarlet Dragon 26‑1 exercise was led and managed by the U.S. Army’s XVIII Airborne Corps at Fort Bragg (Fort Liberty). Within the Army, key executing units and organizations included the 18th Field Artillery Brigade, the 82nd Airborne Division and its Combat Aviation Brigade, and the 108th Air Defense Artillery Brigade. On the technology side, the National Geospatial‑Intelligence Agency (NGA) provided and operated the Maven Smart System, and the Army Program Executive Office for Intelligence, Electronic Warfare & Sensors (PEO IEW&S) Integration Directorate has used recent Scarlet Dragon events as a major venue for integrating its Integrated Sensor Architecture and electronic‑warfare sensors into Maven and corps networks.

Public reporting on Scarlet Dragon 26‑1 notes that “civilian industry partners” participated but does not name specific companies or detail each of their roles. Broader studies of the Scarlet Dragon/Maven Smart System effort show that multiple commercial technology firms were involved over time in building and iterating the MSS AI platform and integrating sensors through the Integrated Sensor Architecture, but which firms were on the ground at this particular 2025 iteration has not been publicly specified. So, beyond saying that tech companies helped provide and test AI‑enabled software, data‑integration tools and possibly drone platforms, the exact list of partners and their roles at Scarlet Dragon 26‑1 cannot be determined from open sources.

The Scarlet Dragon article does not spell out its own safety, legal or ethical rules, but these exercises are governed by existing Defense Department frameworks for responsible AI and weapons testing. At a high level:

• Ethical AI principles: DoD has formally adopted five AI ethical principles—responsible, equitable, traceable, reliable and governable—and directed that AI systems (like Maven Smart System) be designed so humans remain accountable, systems are testable and explainable, and they can be shut off if they behave unexpectedly.
• Testing and safety: New AI‑enabled systems are introduced first in controlled exercises like Scarlet Dragon, described as an “open and minimum‑risk environment,” where performance and failure modes can be observed before fielding.
• Legal compliance: Any use of AI in targeting still has to comply with U.S. law, the law of armed conflict and rules of engagement; current DoD policy keeps humans “in the loop” for lethal decisions and requires review of autonomous weapon functions before they are deployed. Specific checklists or approvals used at Scarlet Dragon 26‑1 are not publicly detailed, but they would sit under these department‑wide responsible‑AI and weapons‑testing policies.