MakinaRocks' illustration of its Runway AI operating system supporting distributed AI across South Korea's military battlefield network.
MakinaRocks’ illustration of its Runway AI operating system supporting distributed AI across South Korea’s military battlefield network. Image: MakinaRocks/LinkedIn

South Korea has demonstrated an artificial intelligence system running on its military battlefield network for the first time during the 2026 Ulchi Freedom Shield (UFS) exercise.

The demo was carried out by South Korean AI company MakinaRocks in cooperation with the Republic of Korea Joint Chiefs of Staff, using its Runway AI operating system within the Korea Joint Command-and-Control System (KJCCS).

Unlike the broader defense network used for administrative and support functions, KJCCS operates on a closed battlefield network with tighter security controls. The demo enabled AI models and services to run directly within that secured environment.

One capability tested was a generative AI-based information retrieval system for searching data held within KJCCS. 

Distributed AI for the Battlefield

MakinaRocks said Runway can support distributed AI deployments connecting data centers, mobile servers, and edge devices in the field.

This architecture could allow AI services to operate across different parts of the network and closer to the battlefield.

The company said the approach will support the development of independent defense AI capabilities by providing an environment for deploying different AI technologies across closed military networks.

You May Also Like

Ukraine’s New AI Platform Spots Patterns and Predicts Enemy Moves

The platform is designed to speed up operational planning by automating battlefield analysis and routine headquarters tasks in a secure environment.

US Army Hires New York’s Andro to Teach AI How to Fight in Dirty Airwaves

The work centers on RF-Gen, a new system that generates synthetic radio frequency data so AI models can train for real-world chaos, such as noisy, jammed, or fully contested airwaves, without dragging expensive field gear into the mix.