RaonSecure Begins Work on Physical AI Security Technology Development... Joint Industry-Academia-Research Project
TECHWORLD ·
✦ AI Summary
RaonSecure has begun developing security technology to build a security safety net for the Physical AI era.
RaonSecure was selected as the lead organization for the government-backed project, "Development of Multi-Layer Security Technology Based on Physical AI for Collaborative Intelligence Systems."
The research focuses on developing step-by-step verification and response technologies against security threats across AI models, data, communications, devices, and operating environments.
RaonSecure has begun developing security technology to build a security safety net for the Physical AI era. The company said it was selected as the lead organization for the government project, "Development of Multi-Layer Security Technology Based on Physical AI for Collaborative Intelligence Systems."
RaonSecure announced on the 15th that it will officially begin research and development. The project receives support from the Ministry of Science and ICT and the National IT Industry Promotion Agency (NIPA).
The project was launched to respond to the possibility that, when Physical AI operates in the real world, cyberattacks or AI judgment errors could cause physical damage. Its purpose is to establish reliability standards for Physical AI to take root in industrial settings.
The research will focus on developing step-by-step verification and response technologies for security threats across all areas of AI models, data, communications, devices, and operating environments. As the lead organization, RaonSecure will oversee the entire R&D effort and will also be responsible for developing identity and access management technology for Physical AI robots and agents.
RaonSecure plans to extend the DID·Decentralized Identity technology applied to the national mobile ID into the Physical AI sector. To that end, it plans to pursue a method of assigning each robot agent a unique "digital ID card." It also plans to establish a system that integrates the management of permissions, delegation relationships, and activity histories.
Once this system is implemented, credentials and permissions can be checked in real time each time a robot performs a mechanical operation. It will also be possible to control commands that fall outside the permitted scope at the execution stage. This extends human identity verification technology into a means of controlling machine behavior, which could make it easier to block hacking and malfunctions before they lead to physical accidents.
The participating companies are dividing responsibilities among their areas of expertise. They are bringing together technical capabilities across AI model verification, encryption, authentication, virtualization, and field demonstration to strengthen the Physical AI security shield.
Each participant is taking on a different role for Physical AI security. SenseTone will develop OTAC(One-Time Authentication Code)-based industrial site access authentication technology and abnormal command detection and blocking technology. OTAC generates a one-time authentication code. Tilon will establish a secure work environment separated from the outside, block external intrusions, prevent information leaks, and detect signs of anomalies inside a closed network.
The companies will also be responsible for using encrypted data and conducting field verification. Disailo will use homomorphic encryption(HE) technology to support analysis and learning on encrypted sensitive information such as process data and AI models. ITstation will connect actual manufacturing and robot-site data and verify whether security technology operates properly in the field.
The academic partners will improve technological reliability and completeness by verifying judgment results and checking stability. They will enhance the reliability and completeness of Physical AI security technology. Sogang University will develop verification technology that proves AI judgment results have not been manipulated while sensitive data and AI model information remain undisclosed. Jeonbuk National University will pre-check the stable operation of AI models even amid environmental changes and cyberattacks, and identify the causes of abnormal signs.
The researchers plan to combine an "intelligent test module" and a "trust integrity module." The "intelligent test module" will be responsible for pre-checking the safety of AI models, while the "trust integrity module" will be responsible for protecting on-site devices in real time. Through this, the researchers aim to implement a standard security platform for Physical AI that can be applied in actual industrial settings.
Through this project, RaonSecure plans to establish an AI and robot security system for smart factory and autonomous manufacturing sites. It then plans to expand the scope of application to various Physical AI industries, including logistics, healthcare, and defense.
Through these efforts, the company expects to secure Physical AI security technology optimized for Korea's industrial environment ahead of time. It also expects to improve technological competitiveness and strengthen its response capabilities in the global Physical AI security market.
RaonSecure CEO Lee Jeong-a said that, in the Physical AI era, the key measure of success is reliability rather than performance. She also explained that the company plans to extend the authentication technology verified through the national mobile ID into the robotics sector. She added that the company intends to establish security standards that will allow AI and machines to be entrusted with confidence in industrial settings.
Source: TECHWORLD · Kim Hye-jin
Original: https://www.epnc.co.kr/news/articleView.html?idxno=408232
References
This article was produced with the help of an automated content generation algorithm.
Source: TECHWORLD
View originalThis article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.