04/08/2026 | Press release | Distributed by Public on 04/08/2026 13:17
By Tina Meketa, University Communications and Marketing
[Courtesy of By Light]
The University of South Florida and defense contractor By Light Professional IT Services LLC have formalized a strategic partnership to advance the development of safer, more reliable artificial intelligence systems for national security and beyond.
The agreement establishes a framework for collaboration across the full spectrum of modern conflict and technology. Key focus areas for future joint initiatives include artificial intelligence, aerospace and space operations, hypersonics, cyber and cybersecurity and the development of next-generation unmanned aerial and ground systems. At the center of the partnership is a shared goal: Ensuring that emerging AI technologies are not only powerful, but also trustworthy.
"As a USF alumnus, I've seen firsthand the university's trajectory in becoming a premier research institution, said Bob Donahue, CEO of By Light. "Formalizing this partnership allows By Light to bridge the gap between academic rigor and operational necessity. USF's world-class research, particularly within the Bellini College of Artificial Intelligence, Cybersecurity and Computing and the Global and National Security Institute, provides the foundational 'AI assurance' required to deploy these technologies in high-stakes environments where failure is not an option."
While By Light specializes in developing advanced simulation environments and AI-driven systems for defense applications, USF researchers bring a distinct strength - using mathematics and theoretical computer science to verify that these systems behave as intended.
"AI systems are already being used to make complex decisions in high-stakes environments. Our role is to ensure those systems are reliable, transparent and aligned with human expectations."
Assistant Professor Ankur Mali
USF Bellini College of Artificial Intelligence, Cybersecurity and Computing
[Courtesy of the Bellini College of Artificial Intelligence, Cybersecurity and Computing]
A key focus of the collaboration is developing what researchers call "AI assurance" - methods that go beyond training algorithms to mathematically evaluate and validate their behavior.
In one project, researchers are working to enhance automated target recognition systems by introducing formal safeguards. These tools are designed to help AI systems distinguish between valid targets and protected entities, while also providing measurable confidence in those decisions.
Rather than relying solely on data and probability, USF researchers are applying formal methods - rigorous mathematical techniques - to establish clear boundaries around how AI systems operate.
"Our work focuses on providing formal guarantees - mathematically proving how these systems will behave under different conditions," said Anshuman Chhabra, assistant professor of computer science and engineering. "It's about adding a layer of robustness and accountability that goes beyond traditional AI development."
Because real-world defense scenarios are difficult - and often impossible to replicate at scale, By Light generates large volumes of synthetic data through advanced simulations. These simulations model a wide range of environments, from varied weather conditions to disrupted communications systems.
[Courtesy of By Light]
USF researchers then analyze this data to ensure the underlying models are sound and that the AI systems trained on it perform reliably across scenarios.
"Since real-world data is limited, simulation plays a critical role," said Seungbae Kim, assistant professor in the Bellini College. "Our goal is to ensure that synthetic data leads to reliable performance in real-world environments."
Another area of collaboration focuses on autonomous systems operating in GPS-denied environments - situations where navigation signals are unavailable or disrupted. Researchers are developing methods to ensure unmanned systems can continue to operate safely and effectively under these conditions, even when external guidance is limited.
While the initial applications are focused on national security, the implications of this work extend far beyond the defense sector.
The same techniques used to validate AI decision-making could be applied in health care, cybersecurity and critical infrastructure - anywhere AI is used to support high-stakes decisions.
The partnership also reinforces USF's role as a leader in emerging technologies and strengthens connections between academia, industry and national security partners. USF's Global and National Security Institute, Institute of Applied Engineering and Bellini College together create an integrated ecosystem to help partners find solutions to national security challenges at the boundary of technology and policy.
By combining real-world applications with foundational research, the collaboration provides opportunities for students and researchers to contribute to cutting-edge innovations with global impact.