Navy Cryptologic Warfare Officer Helps Strengthen Navy Cybersecurity Through Research at Naval Postgraduate School

Executive Summary

U.S. Navy Lt. Kristina Bodeman developed SPELL, an AI-powered system, to identify vulnerabilities in military computer networks. This research, conducted at the Naval Postgraduate School, aims to enhance Navy cybersecurity by enabling local, AI-assisted vulnerability testing for deployed units.

Why It Matters

This research directly addresses critical cybersecurity challenges for military networks, offering a potential solution for identifying vulnerabilities before adversaries exploit them, especially in environments with limited connectivity.

Key Takeaways

• Navy Lt. Kristina Bodeman developed an AI system named SPELL (Spec-to-Shell) to proactively identify weaknesses in military computer networks.
  Source evidence · PDF page 2

  Navy Lt. Kristina Bodeman worked with her thesis partner Swedish Army Lt. Col. Sven Bengt Mattias Gaalnander to develop SPELL, short for Spec-to-Shell, a system that works through user-inputted objectives to successfully gain access to a test computer.

• SPELL is designed to operate locally with smaller AI models, making it suitable for deployed units or ships with unreliable cloud connectivity.
  Source evidence · PDF page 2

  SPELL was designed to explore whether smaller AI models could instead run locally, allowing sensitive information to remain on the system and giving ships or deployed units an option when reliable connectivity is unavailable.

• The system successfully demonstrated its ability to gain access to test computers and guide smaller AI models through complex cyber tasks, proving its potential for AI-assisted cybersecurity.
  Source evidence · PDF page 2

  Using a powerful AI model, SPELL found the file in 37 of 40 runs, confirming completion of the required steps. The team also showed that SPELL could guide a much smaller model through a lengthy, multi-step cyber task, supporting the potential for AI-assisted cybersecurity tools that do not depend on larger, cloud-hosted models.

Strategic Relevance

The development of AI-driven, localized cybersecurity tools like SPELL is strategically important for maintaining network integrity and operational security in contested environments, reducing reliance on vulnerable cloud infrastructure and enhancing the resilience of military systems against sophisticated cyber threats.