Illustration of twin miniature power pylons, one lit, representing a digital cyber twin of a power grid
Use promo code “TCH15” to get 15% off on checkout.

Fujitsu and the Singapore University of Technology and Design (SUTD) are starting a 15-month research project to build cyber twins of power infrastructure, powered by generative AI, so operators can test attacks and defences without going anywhere near a live plant.

Key Takeaways
  • Fujitsu and SUTD’s iTrust are running a 15-month research project to build GenAI-powered cyber twins of power infrastructure.
  • The twins let operators simulate attacks and test defences in a virtual environment, covering IT, OT and the links between them, without risking live plants.
  • The project uses SUTD iTrust’s EPIC testbed and may feed into Fujitsu’s wider Physical AI Security framework.

Testing attacks without touching the real grid

The collaboration, announced on 17 November, pairs Fujitsu with SUTD’s iTrust, the university’s Centre for Research in Cyber Security. The project’s initial focus is the power sector, where operators need to understand whether a defensive measure will hold up without disrupting the physical systems it protects.

A cyber twin is a controlled environment that reproduces how a system behaves, letting researchers run possible attacks and test responses away from live operations. The research will span information technology (IT), operational technology (OT), and the connections between them, using SUTD iTrust’s Electric Power and Intelligent Control (EPIC) testbed for initial development and evaluation.

What each side brings

Fujitsu will contribute multi-agent generative AI workflows for building the twins, plus AI-driven attack simulation and threat-intelligence tooling. SUTD’s iTrust brings its power-sector testbed along with its OT threat research and mitigation expertise. Together, the two sides plan to correlate threat intelligence, simulate attacks against the twins, and test and refine candidate defences.

“Critical infrastructure operators need ways to assess cyber defences without putting essential services at risk,” said Tong Ker Yang, ASEAN CTO at Fujitsu Asia Pacific. “By combining our AI and security research with SUTD’s cyber-physical systems expertise, we aim to develop an evidence-based approach to testing defensive measures and understanding where they can be applied.”

Deciding when a twin is close enough to trust

A core research question is working out measurable criteria for when a cyber twin represents its physical counterpart closely enough to support a real security decision. The team plans to compare results from the twins against physical testbed experiments, and to check whether a mitigation that works on one power-sector system still works on another, and under what conditions.

“Working with Fujitsu, we will investigate how generative AI can support cyber twin development, threat intelligence simulation and how to establish confidence in twin-based testing,” said Mark Goh, Assistant Director at SUTD iTrust. “The goal is to build knowledge and methods that support safer, more resilient systems.”

Fujitsu Research said separately that it would look at folding the project’s findings into its broader Physical AI Security framework. An interview with Tong Ker Yang can be arranged through Fujitsu’s media team.

Author


Discover more from techcoffeehouse.com

Subscribe to get the latest posts sent to your email.

Discover more from techcoffeehouse.com

Subscribe now to keep reading and get access to the full archive.

Continue reading