Machinery Regulation FAQ AI Act overlap

Overlap matters most when a safety function depends on software, sensor data, machine learning, autonomous operation, or fully or partially self-evolving behaviour. The Machinery Regulation does not turn every AI feature into a machinery-specific AI issue; the machinery question is whether the system affects an essential health and safety requirement or the conformity assessment route.

The Commission standardisation request describes the machinery-AI intersection as machinery products with systems ensuring safety functions, with fully or partially self-evolving behaviour using machine learning approaches. CEN-CENELEC Q&A material also frames the issue around predictability: for Machinery Regulation purposes, the concern is relevant when unpredictable or self-evolving behaviour concerns a safety function.

AI-enabled safety functions can affect the conformity route when the machinery or related product falls within Annex I. Products in Annex I Part A are subject to the specific conformity assessment procedures in Article 25(2). Products in Part B use the Article 25(3) route, and self-assessment is possible only where the product is designed and constructed in accordance with relevant harmonised standards or common specifications that are specific to the category and cover all relevant essential requirements.

Article 6 also lets the Commission amend Annex I in light of technical progress, advances in knowledge, or new scientific evidence. One listed criterion for Part A inclusion is uncertainty in existing risk-assessment methods for new machinery categories or technologies, which is the relevant machinery-grounded boundary for novel AI-enabled safety functions.

The Machinery Regulation technical documentation must show how the manufacturer ensures conformity with applicable essential health and safety requirements. For AI-enabled safety functions, the machinery file should be concrete enough to connect the hazard, the protective measure, the safety-related software or data input, and the verification evidence.

Annex IV requires risk-assessment documentation, applied harmonised standards or common specifications, design calculations, test and inspection results, and, where relevant, source code or programming logic of safety-related software after a reasoned authority request. It also calls out sensor-fed, remotely driven, or autonomous machinery where safety-related operations are controlled by sensor data: the file should describe the system's general characteristics, capabilities, limitations, data, development, testing, and validation processes.

The standards context is still a machinery context. The 2025 machinery standardisation request asks CEN and CENELEC to draft or revise standards for machinery products with fully or partially self-evolving behaviour or logic, safety functions governed by such systems using machine learning approaches, and safety functions based on external connections, software, or data that must be protected against corruption.

The request also says standards developed under the machinery mandate should take into account work under the AI Act and Cyber Resilience Act and be prepared as machinery-specific standards. That is useful context for standards monitoring, but it is not a shortcut for claiming compliance with the AI Act.