The EU AI Act Just Banned Emotion AI in the Workplace — Wolfgang Lehmacher Warns Shipping to Pay Attention

Maritime columnist and former NYK head of digital transformation Wolfgang Lehmacher published a sharply timed column in Splash247 on 14 May 2026 warning that the same "emotion AI" technologies now creeping into offices and call centres are starting to appear in maritime safety and crew management pitches — at exactly the moment the European Union has decided to ban them. Emotion AI is the family of systems that claim to read workers' feelings from faces, voices, and keystrokes and translate them into mood, attention, or "attitude" scores. Lehmacher's argument is two-part. The operational case against it: safe operations on a ship depend on crew being willing to say "I'm too tired for this watch," "That manoeuvre doesn't look right," or "We nearly had an accident yesterday." If crew believe a system is scoring their emotional state, they will learn to look calm rather than be honest — and the company gets a clean dashboard and a blinder over what is really happening. The same logic applies to terminal workers, pilots, and shore-based ops teams. The regulatory case: the new EU AI Act explicitly bans systems that attempt to read workers' emotions in the workplace, with narrow exceptions for clearly defined medical or safety use cases. For shipowners, operators, and terminals tied to Europe — which is most of the industry — emotion-tracking is no longer a futuristic ethics debate but a compliance question the legal team needs to answer. Why this matters for ship managers: the EU AI Act's prohibitions on workplace emotion recognition started applying on 2 February 2025, with full enforcement of the high-risk system rules phased in through 2026 and 2027. Several maritime crewing-tech vendors have begun pitching fatigue-detection, attention-tracking, and "wellness" features that sit close to the prohibition line — sometimes on the right side, sometimes not. Lehmacher's practical advice is the right starting point: at your next safety or technology committee, add a single question to every new system review — "Does this product infer emotional state, mood, or attitude from the crew, and if so, where does the EU AI Act draw the line?" AI that detects clear, narrowly-defined safety signals (microsleep, alcohol, falls, distress alarms) with crew involved in the design has a defensible role. Systems that guess emotions to rate performance, decide contracts, or manage behaviour should be a line you do not cross. This is the kind of governance question that does not show up until a vendor demo includes a "crew sentiment dashboard" — by which point your procurement process needs to already know the answer. [Splash247 · EU AI Act, Article 5]

This story is part of the Maritime AI Digest of 17 May 2026.