Maritime Cybersecurity Research Advances With AI and Policy Efforts

Recent research and policy initiatives are addressing cybersecurity challenges in the maritime sector, as digital systems and automation become more widespread in shipping and port operations.

Strahinja Janjusevic, a second-year master’s student at MIT, is conducting research focused on improving the security of maritime infrastructure using artificial intelligence and policy analysis. His work is affiliated with the MIT Institute for Data, Systems, and Society and the Technology and Policy Program.

Janjusevic’s thesis centers on protecting maritime cyber-physical systems, especially older large vessels, by integrating physics-based trajectory modeling with deep learning techniques to identify GPS spoofing attempts. The approach uses a hybrid detection framework that combines an LSTM autoencoder for analyzing signal integrity with a physics-based model that predicts vessel movement based on environmental conditions such as wind and sea state.

During a 2025 internship with Vectra AI’s Network Detection team, Janjusevic examined security risks related to AI agents and the model context protocol, and contributed to a preprint on using these technologies for agentic red teaming activities. He is also active in the MIT Maritime Consortium, which includes partners from academia, industry, and regulatory bodies in countries such as Singapore and South Korea.

What the numbers show

• A 2025 systematic review highlighted the effectiveness of AI-driven cybersecurity tools in maritime networks
• Studies in 2025 documented mariners’ experiences with GPS spoofing and ransomware attacks
• Two conferences on technology and security are scheduled for February and April 2026

Maritime systems are facing increased vulnerability as information technology and operational technology systems become more interconnected, and as ships adopt satellite communications, autonomous navigation, and sensor networks. This expansion of digital infrastructure has led to a larger attack surface for potential cyber threats.

Research published in 2025 identified that AI-based methods such as machine-learning intrusion detection, anomaly detection, and predictive threat modeling can improve real-time identification of threats in maritime networks. However, these studies also noted ongoing challenges, including limited data availability, regulatory issues, and the risk of adversarial attacks targeting AI systems.

A separate 2025 study involving mariners reported firsthand encounters with GPS spoofing and ransomware incidents. The study also found that there are gaps in cybersecurity training, detection tools, and overall awareness within the maritime sector.

Janjusevic is involved in organizing upcoming conferences focused on European policy and technology’s role in national security, scheduled for early and mid-2026 and co-hosted by Harvard and MIT. These events aim to bring together stakeholders from different sectors to discuss ongoing developments and challenges in maritime cybersecurity.

* This article is based on publicly available information at the time of writing.