DNV and Havtil Deploy AI Tool Mining Decades of Offshore Safety Data

DNV has partnered with the Norwegian Ocean Industry Authority (Havtil) to develop an AI-powered search and analysis tool that mines decades of accumulated offshore safety data to identify patterns and predict risk factors, with initial focus on reducing danger from dropped objects—one of the persistent hazards across maritime and offshore operations. The collaboration addresses a fundamental challenge facing the industry: Havtil has collected extensive high-quality data through inspections, investigations, and damage analyses over several decades, but this information remains difficult to access for systematic learning. The DNV report, titled "Use of Artificial Intelligence to Reduce Danger and Risk of Accidents Due to Falling Objects," demonstrates how AI can transform buried institutional knowledge into actionable safety intelligence. Ontology-based knowledge extraction: The project developed a specialized ontology for falling objects—a structured vocabulary defining relationships between incident causes, contributing factors, and outcomes—enabling AI systems to understand the complexity and nuance of human-written investigation reports. Morten Langøy, Principal Engineer at Havtil, noted the collaboration represents "a promising step forward for using AI on industry safety data," with plans to expand the ontology together with industry domain experts to create a more comprehensive model for incident causation. Natural language query capability: The AI system enables safety professionals to ask practical questions through natural language interfaces, such as "What are the most common causes of falling objects during lifting operations on platform X?" or "Is there a correlation between the time of day and the severity of falling object incidents?" or "What were the common contributing factors in dropped object incidents in the last year?" The system analyzes historical data and synthesizes results that would previously require manual review of thousands of documents by specialized investigators. Industry-wide potential: The report highlights transformative possibilities if the system could access anonymized data gathered from across the entire industry. Pattern recognition across multiple operators and installations could reveal systemic risks invisible within individual company datasets, revolutionizing safety practices by identifying emerging hazards before incidents occur. DNV's two decades of ontology engineering experience, combined with advances in large language models, positions this technology to fundamentally change how the maritime and offshore industries learn from past incidents to prevent future accidents. [DNV]

This story is part of the Maritime AI Digest of 30 November 2025.