2 December 2026: the AI deadline most shipping firms haven't noticed

The EU pushed back its toughest AI rules. It did not push back all of them, and the reach extends to London.

When the EU's Digital Omnibus on AI came into force on 27 July 2026, much of the coverage read like a reprieve. The strictest obligations for high-risk AI moved back by more than a year. Many firms took that as permission to stop thinking about it.

That is a mistake, for three reasons.

1. One deadline arrives on 2 December 2026

Under Article 50 of the AI Act, AI-generated content must be marked in a machine-readable way. For systems placed on the EU market before 2 August 2026, the grace period ends on 2 December 2026. Systems launched after that date had to comply from day one. The duty to tell people they are dealing with an AI system has applied since 2 August.

If your business uses AI to generate images, reports, chat replies or marketing content for EU audiences, check how that output is labelled now.

2. The delay is a delay, not a cancellation

Stand-alone high-risk systems, listed in Annex III, now apply from 2 December 2027. AI built into products covered by EU safety law, in Annex I, follows on 2 August 2028.

Annex III includes AI used in recruitment and in managing workers. For shipping, that means crewing platforms that screen seafarer CVs or rank candidates are squarely in scope. Thirteen months is not long to document training data, set up human oversight and prove the system is accurate and fair.

3. Being in London does not take you out of scope.

The AI Act applies to providers who place AI on the EU market and to providers and users outside the EU whose AI output is used inside it. A UK ship manager serving EU owners, or a London broker using AI in EU-facing work, can be caught.

The governance gap

The industry is enthusiastic about AI but thin on control. In a 2026 Thetius and Marcura survey of maritime professionals, 81% said they were running AI pilots. Only 11% had formal policies for scaling AI, and only 17% had transparent processes for how AI makes decisions.

That gap is exactly what regulators, charterers and insurers will ask about. The Omnibus also softened the AI literacy duty: firms must take measures to support staff literacy rather than guarantee a set level. That is easier to meet, but it is still a duty.

What to do this quarter

List every AI system you build, buy or use, and who it affects.

Mark which ones produce content seen in the EU, and fix labelling before 2 December.

Flag anything touching recruitment, crewing or staff management as likely high-risk.

Write a short AI policy: who approves new tools, how outputs are checked, and who is accountable.

Put a human in the loop wherever a decision affects safety, money or people.

The takeaway

The Omnibus bought time. It did not buy a pass. The firms that use the next year to build governance into their AI, rather than bolt it on later, will move faster when the deadlines arrive.

vi-tech designs and builds AI systems with oversight, logging and documentation from the start for maritime clients and others across the UK and beyond. This article is general information, not legal advice.