In early 2026, Rosa Szurgot, assistant professor of Cyber Intelligence and Security, presented OrbitWhisperer, an AI-driven satellite-resilience framework, to NATO’s Science and Technology Office in Riga, Latvia.
Szurgot began to develop the technology after identifying a critical gap in satellite cybersecurity.
“Modern satellites operate in an environment where cyber, physical and radio-frequency threats intersect in ways we’ve never experienced before,” she explains. “Most current space systems still rely on legacy firmware, limited onboard computing and assumptions that the space environment remains ‘benign.’ ” After determining the issue, Szurgot felt that implementing AI capabilities was the clear solution.
“OrbitWhisperer emerged from a simple question: What if a satellite could learn from its own telemetry, recognize early warning signs and protect itself in real time?” she says. “That capability — resilient autonomy — felt both urgently needed and technically achievable with the right blend of AI and cybersecurity.”
How OrbitWhisperer Works
OrbitWhisperer implements two types of technology currently used in satellite cybersecurity: fuzzing and machine learning. However, Szurgot’s technology takes a different approach than typical methods.
“Traditionally, fuzzing and machine learning are used separately — fuzzing uncovers software vulnerabilities, while machine learning analyzes telemetry or communication patterns. OrbitWhisperer blends the two,” she explains.
“We treat fuzzing not just as a testing method, but as a generator of synthetic satellite failures that train our models to recognize real anomalies. This helps overcome a major challenge in space cybersecurity: the scarcity
of labeled failure data. OrbitWhisperer essentially produces a system that is more adaptable, less dependent on handcrafted signatures and better suited for the unpredictable nature of space operations.”
Making International Strides with NATO
OrbitWhisperer was selected through NATO’s Science and Technology Organization (STO) Technical Activities process, which helps identify research that has potential implications for allied defense. Szurgot’s team was invited to present its research at NATO’s offices in Riga.
“The reception was extremely positive. Many attendees expressed that the integration of fuzzing, AI and cybersecurity into one cohesive satellite-resilience framework filled a gap they had been grappling with in their own work,” she says. “Several delegations were interested in exploring how OrbitWhisperer might translate into their operational environments.”
Looking Ahead for OrbitWhisperer
For OrbitWhisperer’s future, Szurgot foresees refining its technology, continuing collaboration with NATO and preparing the research for journal publication.
“The work presented at the NATO meeting in Riga serves as the foundation for that publication — this presentation paper is the first formal step toward making OrbitWhisperer accessible to the broader research community,” she says.
Longer-term, she hopes to integrate OrbitWhisperer into digital-twin environments to evaluate its performance under realistic orbital dynamics and operational constraints.
“The ultimate vision is to develop a lightweight, deployable module that satellite operators can incorporate directly into their architectures to enhance autonomy, anomaly detection and cyber resilience,” she explains
Realizing that vision will require strong partnerships between Embry-Riddle’s Center for Aerospace Resilient Systems (CARS) and key partners around the world.
“From an Embry-Riddle standpoint, we hope to bring OrbitWhisperer into defense-relevant test environments and begin active collaboration with international partners responsible for satellite operations and space security,” Szurgot says. “This is a natural evolution of the project: moving from academic research into applied, operationally meaningful experimentation.”