AI governance in manufacturing.

EU AI Act

AI embedded in machinery, vehicles, and industrial equipment regulated under existing EU product safety legislation — the Machinery Regulation, the Automotive Regulation, the Radio Equipment Directive — is subject to the EU AI Act's Annex I classification. This creates a cumulative compliance obligation: product safety regulation plus AI Act requirements.

Product safety and CE marking

AI-enabled machinery placed on the EU market must carry CE marking that encompasses both the product safety requirements and the AI Act obligations where applicable. The interaction between the two regimes requires careful legal analysis for each product category.

ISO 13849 / IEC 62061

Functional safety standards for machinery require that AI components in safety-related control systems be assessed against reliability and integrity requirements. Traditional software safety assessment methodologies do not translate directly to machine learning systems.

Environmental and supply chain regulation

AI used for environmental reporting, emissions monitoring, and supply chain traceability is increasingly subject to regulatory requirements including the EU Corporate Sustainability Reporting Directive. AI-generated regulatory data must be accurate and auditable.

Quality control AI without validation for edge cases

Manufacturing quality AI trained on standard production runs performs well on average but may miss rare defect types that occur at the tail of the distribution — exactly the cases that matter most in safety-critical applications. Validation must explicitly test tail performance.

Predictive maintenance AI and distributional shift

Predictive maintenance models trained on historical equipment data may fail when equipment ages, is modified, or operates in different environmental conditions than the training data. Without active monitoring, model degradation goes undetected until equipment failure occurs.

AI governance gaps in supply chain AI

Manufacturers that incorporate AI-enabled components from suppliers inherit governance obligations for those components. AI in embedded systems, sensor fusion, and control units requires supply chain AI governance — not just product safety assessment.

Workforce AI without consultation

AI-assisted production scheduling, performance monitoring, and workforce management tools deployed without adequate worker consultation and governance have attracted regulatory attention and industrial action in multiple jurisdictions.

Have you assessed which AI systems embedded in your products fall within the EU AI Act's Annex I classification — and how that interacts with existing CE marking obligations?

For AI in safety-critical applications, what functional safety assessment has been conducted and against which standard?

How is AI component governance addressed in your supplier qualification process — specifically, what governance evidence do you require from AI technology suppliers?

What monitoring do you have in production for AI quality control and predictive maintenance systems — and what thresholds trigger human review?

Have you assessed the product liability implications of AI components in your products under the EU Product Liability Directive and applicable national law?

What is your process for AI incident reporting for products already in the field — and who has responsibility for coordinating that process?