//

The Trust Gap in Maritime AI

The Trust Gap in Maritime AI

Maritime already runs on AI every day. Trusting it to act is the harder part. New research from Thetius and Marcura, free to download.

Maritime already runs on AI every day. Trusting it to act is the harder part. New research from Thetius and Marcura, free to download.

By submitting I agree that Marcura may use my details to respond to this request. See our Privacy Policy.

In this report

  • Where AI already sits in daily work: How far adoption has gone, and the gap between using AI and being ready to rely on it.

  • The verification tax: The time spent checking AI before anyone acts on it, and how to tell warranted checking from the kind that falls as confidence grows.

  • From assistant to agent: What changes as AI moves from suggesting to doing, and why governance has not kept pace.

  • From trust to accountability: Where a person stays on the decision, and the five priorities for putting AI to work with confidence.

Maritime adopted AI faster than it learned to trust it

A new study from maritime research house Thetius and technology company Marcura looks at where that trust sits today: how far AI has spread into daily work, how much of the day now goes on checking it, what warrants human scrutiny and what does not, and how the companies furthest ahead are turning verification from a cost into a source of learning.

With contributions from operators, technologists and researchers across the industry, including Tomini Chartering.

What the industry told us

“There’s a lack of understanding of what AI can actually do. There’s no formal training on it, so most AI use is trial and error by individual managers rather than an approved method.”

Jack Kitching, Master and Operations Manager, Islands Unlimited

“People want to apply technology everywhere rather than asking how the problem can be solved. You need to fall in love with the problem.”

Manuel Fernandez de Mosteyrin, Stena Bulk

“It’s a human-to-AI-to-human process: the human puts in the knowledge, the AI curates and connects it to the clause in front of you, and the human makes the final decision.”

Thomas Stjernholm, Managing Director, Commercial, Tomini Chartering

Frequently asked questions

How is AI being used in maritime today?

AI is now part of everyday work in maritime. In the research, 63% of professionals said they use AI every day. It shows up most in tasks that involve working through large amounts of information and making repeated commercial judgements: reviewing charter parties against a company's own history, matching purchase orders, invoices and receipts in procurement, drafting and organising correspondence, and preparing claims. Most of it still runs on general tools rather than maritime-built ones. 42% said their organisation gets its AI capability mainly from general providers such as ChatGPT or Copilot, against 10% from specialist maritime vendors.

Does AI actually save time in shipping, given all the checking?

Yes, on balance. 85% said AI saves them time overall once the checking and correction are taken into account, even though 55% spend an hour or more a week checking or correcting its outputs. The report calls that checking the verification tax: the time and attention spent verifying AI-generated work before anyone acts on it. Some of it is warranted by the stakes of the decision and stays. The rest is driven by low confidence, and it falls as AI proves itself.

Can you trust AI outputs in shipping?

Most maritime professionals do not fully trust them yet. Asked how far they trust what AI produces, 57% placed themselves in the middle of a five-point scale, neither distrusting it nor fully trusting it. The caution is earned: 27% find errors often or almost always when they check, and few describe their organisation's data as fully ready for AI to rely on. Trust tends to grow when people can see the results the system reached and how it reached them.

Who is accountable when AI makes a commercial decision at sea?

A named person, according to most of the industry. 80% said a named person should always be accountable for a commercial decision, however capable AI becomes. For important business decisions, 70% would always check an AI recommendation before acting on it. The pattern the report describes is human-to-AI-to-human: AI organises the information and surfaces the risks, and a person makes the call and owns it.

Is the maritime industry ready for agentic AI, the kind that takes action?

Adoption is ahead of readiness. 72% said their organisation is already using, piloting or planning agentic AI that can take action rather than only make recommendations, such as drafting a handover between the chartering and operations desks. Yet only 15% said their governance is ready for AI that acts. The report's advice is to grow that authority task by task, matching how much a system is allowed to do to the consequences of getting it wrong.

How should a shipping company govern AI, and does the EU AI Act apply?

Governance is lagging adoption, and closing that gap is the priority. 35% said their organisation now has a formal AI policy, up from 11% a year earlier, but only 25% said it has one that people actually follow. Regulation is arriving alongside: the EU AI Act became broadly applicable on 2 August 2026, with its AI literacy requirements in force since February 2025. The report's guidance is to fit AI into existing risk and compliance structures rather than build a separate process for every tool, keep a record of which systems are used and what they can access, and match human oversight to the risk each decision carries.

63%

use AI every day

57%

don’t fully trust the AI outputs they rely on

27%

find errors often or almost always when they check

15%

say their governance is ready for AI that takes action

Source: Thetius and Marcura, Earning Trust: AI in Maritime (2026).

Source: Thetius and Marcura, Earning Trust: AI in Maritime (2026).

By submitting I agree that Marcura may use my details to respond to this request. See our Privacy Policy.

Get the report

By submitting I agree that Marcura may use my details to respond to this request. See our Privacy Policy.