South Korea’s RaonSecure is launching the development of security technology designed to prevent hacking and malfunctions by assigning unique digital IDs to industrial robots and artificial intelligence (AI) agents. The core objective is to establish a system that verifies the identity and authorization of AI and robots before they execute commands in manufacturing environments, effectively blocking unauthorized operations at the execution stage.
RaonSecure announced on the 15th that it has been selected as the lead agency for the “Development of Physical AI-based Multi-layered Security Technology for Collaborative Intelligence Systems,” a project supported by the South Korean Ministry of Science and ICT and the National IT Industry Promotion Agency (NIPA). “Physical AI” refers to technology that allows robots, logistics equipment, and other physical devices to perceive, judge, and control their own movements. Because cyberattacks or AI errors in these systems can lead to more than just data breaches—potentially causing equipment malfunctions, production line shutdowns, and safety accidents—a specialized security framework is essential.
The project focuses on developing technology to verify and respond to threats at every stage of the system, including AI models, data, communication networks, hardware, and operating environments. In smart factory and autonomous manufacturing settings where multiple robots and AI agents exchange information to collaborate, the system will continuously verify who issued a command, whether they have the authority to execute it, and whether the results have been tampered with.
RaonSecure plans to extend its Decentralized Identity (DID) technology, previously applied to South Korea’s national mobile identification cards, to the realm of robots and AI agents. By issuing unique digital IDs to each device, the company intends to manage device-specific permissions, delegation relationships, and task histories centrally. Every time a robot operates equipment, the system will verify the authentication credentials and scope of authorization, blocking any commands that fall outside of registered permissions before they are carried out.
Lee Soon-hyung, CEO of RaonSecure, stated, “The success of the Physical AI era depends not on how advanced the technology is, but on how trustworthy it is. By extending the authentication technology proven through our national mobile ID system to the robotics sector, we will establish security standards that allow industrial sites to safely entrust tasks to AI and machines.”
In addition to RaonSecure, several security firms and universities are participating in the project to build a multi-layered defense system. Senstone is responsible for access authentication using One-Time Authentication Codes (OTAC) and technology to detect and block abnormal commands. Tilon is developing secure work environments isolated from external networks and technology to detect anomalies in closed networks.
Desilo is applying homomorphic encryption technology to enable analysis and learning while keeping sensitive information encrypted. IT Station will link manufacturing and robotics field data to validate the technology. Sogang University is tasked with developing technology to prove the integrity of AI decision-making results without exposing sensitive information, while Jeonbuk National University will handle technology for assessing the stability of AI models under environmental changes and cyberattacks, as well as analyzing the root causes of anomalies.
As the autonomy of robots in industrial settings grows, the focus of security is expected to shift from blocking network intrusions to controlling the actions of devices and AI. In environments where production facilities and logistics systems are interconnected, a single authentication error or malicious command can lead to a chain reaction of work stoppages, making pre-execution verification more critical than post-incident response.
RaonSecure plans to establish a security foundation for smart factories and autonomous manufacturing through this research, with the goal of expanding its application to other Physical AI industries such as logistics, healthcare, and defense. However, for full-scale commercialization, the company must still overcome challenges related to performance verification, specifically minimizing authentication latency while accurately identifying abnormal behavior across diverse manufacturing environments.
Click Here For The Original Source.
——————————————————–
