AI Security Measures Target Electric Vehicle Battery Risks
A new federal initiative aims to protect electric vehicles from cyberattacks by hardening the AI systems that manage their batteries.
Electric vehicle batteries are increasingly managed by artificial intelligence, creating a new target for cyberattacks that could disrupt safety and grid stability. To address this, a research team led by the University of North Dakota has secured funding to develop defensive AI models that can detect and neutralize malicious manipulations in battery management systems.
The project, selected from over 5,000 applications for the U.S. Department of Energy's Genesis Mission, includes researchers from La Sierra University, the University of Wisconsin-Madison, and Clean Republic SODO. The primary goal is to ensure that the AI controlling critical energy infrastructure remains resilient against adversarial scenarios where data might be compromised or manipulated by bad actors.
Securing Critical Energy Infrastructure
As AI becomes more embedded in critical energy systems, the risk of sophisticated cyberattacks grows. The research team aims to build an adversarial robustness framework that allows battery management systems to identify when their inputs have been tampered with. This is crucial because a successful attack could lead to significant safety hazards or reliability failures in millions of electric vehicles on the road.
Overcoming Data Scarcity Challenges
A major hurdle for the team is the limited availability of real-world data involving cyberattacks on battery systems. Without enough examples of compromised systems, AI models cannot learn to recognize and defend against these specific threats. Consequently, the researchers are developing methods to generate high-quality synthetic data to expand their training sets, which is essential for improving the accuracy of their defensive algorithms.
Collaborative Academic Effort
La Sierra University serves as a key partner in this initiative, with Assistant Professor Fuhao Li acting as the principal investigator. The institution is one of only 13 higher education organizations in California participating in the DOE Genesis Mission. This collaboration highlights how academic institutions are working together to solve complex technical problems that affect national energy security and public safety.
The initial phase of the project, supported by the Genesis Mission award, focuses on testing these AI models in simulated environments. While the immediate work is theoretical, the long-term objective is to apply these defenses to real battery systems and potentially extend them to smart grid environments, ensuring that the growing network of electric vehicles remains secure and trustworthy.
