DOE Funds U of A Research to Safeguard the AI Behind America's Electric Grid

America's electric grid increasingly relies on artificial intelligence to detect problems, respond to disruptions and help keep electricity flowing to homes and businesses. But those same AI systems can become targets for cyberattacks that cause them to misread data and make costly mistakes.

To help address that threat, Qinghua Li, professor of electrical engineering and computer science at the U of A, received a $750,000 award from the U.S. Department of Energy to develop new tools that can identify and defend against attacks on AI-powered grid operations.

"I am excited!" Li said. "It's good to see our innovative idea is well received in this highly competitive nationwide competition, and to see the U of A be a part of this Genesis Mission."

The Genesis Mission is a nationwide effort launched by executive order in 2025 and led by the Department of Energy to harness AI and supercomputing to advance American scientific discovery and strengthen national security. Li's project, "An Autonomous Framework for Adversarial Robustness and Resilience of AI Models in Power Grids," is part of this effort.

AI plays an important role in detecting faults, classifying grid events and making split-second decisions that keep electricity flowing in a power grid. However, hackers can create manipulations to the system, known as adversarial attacks, that cause AI models to misread data, misidentify problems and trigger incorrect responses. In a power grid, even the smallest error in protections can cascade into widespread outages.

Li and his team aim to close that gap.

This project will develop an autonomous AI toolkit capable of discovering and defending against hidden vulnerabilities in the AI models that support power grid operations. Unlike most existing research, which focuses on single models in fields like image recognition and natural language processing, this toolkit will examine entire workflows consisting of connected AI models.

Once officially awarded, the project is expected to accelerate vulnerability detection, protect AI-enabled grid operations and contribute to the security of the nation's next-generation energy infrastructure. The project will also support student training and development across engineering disciplines.

Xintao Wu, professor of electrical engineering and computer science at the U of A, is a co-investigator of this project. Brookhaven National Laboratory and Southern Methodist University are also partnering with the U of A on this effort.

About the College of Engineering: The University of Arkansas College of Engineering is the state's largest engineering school, offering graduate and undergraduate degrees, online studies and interdisciplinary programs. It enrolls more than 4,700 students and employs more than 150 faculty and researchers along with nearly 200 staff members. Its research enterprise generated $47 million in new research awards in Fiscal Year 2025. The college's strategic plan, Vision 2035, seeks to build the premier STEM workforce in accordance with three key objectives: Initiating lifelong student success, generating transformational and relevant knowledge, and becoming the destination of choice among educators, students, staff, industry, alumni and the community. As part of this, the college is increasing graduates and research productivity to expand its footprint as an entrepreneurial engineering platform serving Arkansas and the world. The college embraces its pivotal role in driving economic growth, fueling innovation and educating the next generation of engineers, computer scientists and data scientists to address current and future societal challenges.