Marine insurer NorthStandard and AI navigation firm Orca AI have published joint operational data covering 10.8 million nautical miles of container voyages, finding that high-severity close encounters fell by 52% across the cohort within a year of Orca AI's deployment. The study is being positioned as the first time a P&I club and an AI navigation provider have combined fleet-scale data to quantify how the technology affects real-world bridge behaviour.
The 139-vessel cohort showed a 22% reduction in high-severity encounters within the first six months, doubling over the second half of the year. Severity was measured against four objective parameters: Closest Point of Approach, Time to Closest Point of Approach, COLREGs interactions, and traffic density. Improvements held across vessel ages, which the report argues suggests that AI-assisted situational awareness can lift safety performance regardless of how modern the underlying bridge equipment is.
Geographic breakdowns are sharper in congested waters. The North and Baltic Seas posted a 36% reduction in high-severity encounters; the China Sea and Japan Sea, 18%. Crew usage of the Orca AI platform also rose in those zones, which the report reads as evidence that bridge teams lean on the system most heavily when traffic density and decision pressure are highest.
What this joint analysis does is validate, at scale, what we've been seeing across our customer base for several years: that earlier, better-informed decisions on the bridge lead directly to safer voyages. As this shift becomes measurable and consistent across fleets, we can expect it to increasingly be reflected in reduced risk exposure, and over time, in how insurers assess that risk.
Fleet growth, crewing shortages, rising asset values, and increasing disruption to navigation systems, including GNSS interference and spoofing that recent geopolitical conflicts have exposed at scale, are compounding navigational risk. The operating environment today is more complex, less predictable, and less forgiving. What we're seeing through our work with Orca AI, and now reinforced by this study, is that improved situational awareness and earlier risk detection can materially reduce close-quarters situations.
Two notes on what the study does and does not show. It is a vendor-and-insurer joint analysis rather than independent academic research, and the 139-ship sample, while large in absolute terms, is concentrated in container shipping. The headline 52% reflects the difference between a three-month adaptation phase and a months 10 to 12 stabilised-usage phase on the same vessels, not a comparison with a control group. That said, the methodology is unusual in being grounded in objective COLREGs and CPA data rather than incident reporting, and the consistency across vessel ages and trade lanes is the more striking result.
Orca AI's platform is now installed on more than 1,000 vessels worldwide. NorthStandard provides cover for over 365 million GT of owned and chartered tonnage across the merged North P&I Club and Standard Club book.
