Maritime is embracing AI with great enthusiasm, incorporating it into an increasing number of workflows to speed up mammoth tasks such as just in time arrival, data categorisation, voyage optimisation, vessel design, and more. While some of the benefits are undeniable, particularly those that have been carefully calibrated to match the capabilities of the AI systems chosen, a new risk has emerged in the form of geopolitical impact.

The recent decision by the American government to restrict access to Anthropic’s Fable 5 and Mythos 5 model for users outside the US serves as a vital case study in risk management as underscores an uncomfortable operational reality: many proprietary maritime systems are built upon third-party foundational models that remain subject to external geopolitical interference.

While many may argue that maritime’s technology needs are bespoke to their operations, the fact is that the software architecture of these solutions rarely relies on bespoke AI models built from the ground up. Instead, technology vendors and in-house development teams typically utilise digital infrastructure (including servers) provided by a small number of technology providers based primarily in the US or China despite the fact that they may be branded to appeal to local markets or specific sectors.

When a sovereign state implements export controls or data-sharing restrictions, the downstream effects on global supply chains can be immediate. For a ship owner or manager, the sudden decoupling of a management system from its underlying AI framework presents a severe operational risk. Without the baseline model, the proprietary application layer cannot function correctly, effectively rendering a substantial capital investment obsolete and creating a stranded asset. This can be extremely problematic when critical software becomes unusable, as it could disrupt fleet compliance, maintenance forecasting, and voyage efficiency.

Consequently, procurement and risk-management criteria for maritime technology must evolve. Traditional software vetting focuses primarily on data security, return on investment (ROI), and user interface integration. Moving forward, a key metric in the corporate due diligence process must be the geographical and political resilience of the foundational AI architecture. Chief Technology Officers will need to vet the jurisdictional alignment of their software and services value chain, model redundancies and portability of information  – and in some cases, there may even be justification to look at sovereign systems with on-premise deployment.

While this may sound like a mammoth task, we must remember that this is an issue that will also be impacting other sectors in any country, and thus we are not alone in our exposure. The flip side of this coin is that we also have the potential to mitigate the risks via collaborative cross-sectoral demand creation and investment in digital infrastructure. In fact, a number of government-level conversations that I am lucky enough to be privy to have highlighted the desire to cultivate ‘home-grown’ AI infrastructure and services – with serious money being brought to the table. This is particularly true at a time that deliberate interference with data, skewing of outputs, and surveillance of information and plans is a legitimate consideration for many countries’ national security.

As maritime continues to integrate machine learning into critical decision-making processes, ensuring the continuity of these systems is paramount. True digital resilience requires a clear understanding of the software supply chain secure and geopolitically stable digital and physical assets. I truly hope that our sector doesn’t opt to focus on the fairytale of AI without considering the bogeyman of geopolitical interference that lurks under our collective beds.

———–

Written by Namrata Nadkarni, CEO of Intent Communications. Image created by Gemini.
————-
Click here to sign up for our free monthly ‘De-Mystifying Digitalisation and AI in Maritime’ newsletter and get the latest cross-sector expert insights, news, interviews, videos and case studies straight to your inbox.