EU Machinery Regulation Machinery Regulation vs EU AI Act

The EU AI Act creates a high-risk category for AI used as a safety component of a product covered by EU harmonisation law (Annex I) when that product requires third-party conformity assessment for placing on the market or putting into service (Article 6(1)). For machinery, this often maps to product families that fall into notified-body routes under the Machinery Regulation (Annex I + Article 25).

In other words: if your "AI-enabled safety function" is part of a CE-marked machinery safety case and you need a third-party route for the product category, treat the AI as a high-risk AI system and plan to meet both evidence expectations.

For integrated products, the AI Act pushes responsibilities up the chain. In the case of high-risk AI systems that are safety components of products covered by certain EU harmonisation legislation, the product manufacturer is considered the provider of the high-risk AI system in specified circumstances (Article 25(3)).

The AI Act also treats certain re-branding, intended-purpose changes, or substantial modifications as provider triggers (Article 25(1)). This matters for machinery manufacturers who fine-tune models, swap perception stacks, or change intended use (for example from "operator assistance" to "operator protection").

A smart way to avoid duplicate work is to align your AI Act high-risk evidence with the Machinery Regulation's technical documentation and risk assessment structure. The AI Act requires a lifecycle risk management system (Article 9), data and data governance controls for trained systems (Article 10), and provider duties including documentation keeping, log retention, conformity assessment, declaration of conformity and CE marking for the high-risk AI system (Article 16).

The Machinery Regulation technical file expects a coherent safety case: risk assessment, protective measures, residual risk rationale, tests/verification, and (when relevant) software and control-system evidence (Annex IV). For AI-driven safety functions, make the mapping explicit: hazard -> AI failure mode -> mitigation -> verification -> monitoring trigger.

The Machinery Regulation route is selected via Annex I category logic and Article 25 modules; some categories require third-party involvement. The AI Act requires providers of high-risk AI systems to ensure the relevant conformity assessment procedure is completed before placing on the market or putting into service, and to draw up declarations and affix CE marking for the high-risk AI system (Article 16).

For AI as a safety component in machinery, you want a single combined technical narrative: a product safety case with an embedded AI safety case. Practically, this means one cross-referenced evidence index, consistent versioning, and a change log that both machinery safety reviewers and AI compliance reviewers can follow.

The two frameworks phase in on different schedules. The AI Act entered into force on 1 August 2024 and applies on a staggered timetable, while the Machinery Regulation entered into force on 19 July 2023 and applies generally from 14 January 2027.

If you ship AI-enabled machinery, run one roadmap: machinery route selection, software evidence, and technical-file modernization first, then AI Act high-risk controls layered into the same governed evidence system.