If you design or integrate machinery, you probably used to ask: “Which standard do I need to follow to be compliant?” That used to be a straightforward question. Not anymore.
With the new EU Machinery Regulation (2023/1230) coming into effect on January 20, 2027, that simple mindset is outdated. This Regulation replaces the long-standing Machinery Directive (2006/42/EC) and shifts the focus from just functional safety to something broader: resilience. In short, machines now need to be not just safe – but also secure and trustworthy.
“Why now”: Regulation 2023/1230 widens the risk lens
The new Regulation significantly expands the Essential Health and Safety Requirements (EHSRs) to include digital risks – changing how we design, validate, and maintain machinery.
1. Cybersecurity and protection against corruption.
For the first time, EHSRs explicitly require the protection of safety‑critical software and data, and demand that connectivity (local or remote) cannot lead to hazardous situations. In Annex III 1.1.9, software/data critical for compliance must be identified and adequately protected against accidental or intentional corruption – shifting the assessment from “what if it fails?” to “what if someone makes it fail?”
2. AI and high‑risk systems.
If your machinery includes AI – especially if it adapts or acts autonomously – you will need to evaluate how its behavior could evolve over time, especially in interaction with humans. This ties directly into the EU AI Act, which requires high-risk AI systems to meet strict standards for accuracy, robustness, and cybersecurity.
3. Digitalization and the documentation lifecycle.
The Regulation also embraces digitalization: instructions and technical documentation can now be digital-only, as long as they are accessible, printable, and paper copies are available on request (and mandatory for non-professional users). Your technical file and risk analysis are not static anymore – they must be maintained throughout the lifecycle as threats and software/AI behavior evolve.
How to show conformity: Harmonized standards and State‑of‑the‑Art
The legal lever remains the presumption of conformity when you design to the relevant harmonized standards (hEN) under the Regulation. For control systems, core references remain EN ISO 13849‑1 (PL) and EN IEC 62061 (SIL); ISO 25119 (AgPL) applies in agriculture. Note that EN ISO 13849‑1 was revised in 2023, and the EU has launched gap analyses and a standardisation request to update the hEN landscape for the new digital EHSRs. Don’t expect the paper trail to save you if the risk is new: today, there is a gap, and you must bridge it.
Beyond hEN, use state-of-the-art standards to support your engineering decisions – especially for AI-driven or software-heavy systems:
- IEC 61508: Core framework for safety lifecycle.
- IEC 62443: Industrial cybersecurity.
- ISO/IEC TR 5469 and TR 24028: Address AI behavior, trustworthiness, and robustness.
- ISO/IEC 42001: AI management systems – showing how you govern AI across its lifecycle.
Two Structural Shifts to Watch
- Annex I now consolidates high-risk machinery categories – many of which will require third-party assessment. Keep an eye on changes through delegated acts.
- If you make a substantial modification – like a software or ML model update – you become the manufacturer under the law, with full legal responsibility.
Why This Is Urgent
Standards bodies like CEN/CENELEC are still updating harmonized standards to reflect new cybersecurity and AI risks. In the meantime, simply checking the old compliance boxes (EN ISO 13849, EN IEC 62061) isn’t enough. You need to anticipate and address these risks now.
Three strategic priorities for 2027
1. Redefine risk: Cyber & AI are safety risks.
Expand your risk assessments (like ISO 12100, FMEA, or HARA) to include threats like data corruption, sensor manipulation, and erratic AI behavior. Link mitigations to both safety and security measures, and clearly document assumptions and residual risk.
2. Manage the digital lifecycle: compliance is continuous.
Set up systems to:
- Keep technical documentation current and secure
- Manage software updates and vulnerabilities responsibly
- Re-assess safety each time software or AI changes
This also aligns with future obligations under the AI Act and the Cyber Resilience Act (CRA).
3. Integrate Safety, Cyber, and AI Functions.
Your functional safety, cybersecurity, and AI teams must collaborate on a unified risk model and change management process. That is also your best bet for earning presumption of conformity once harmonized standards are updated.
Final thought
The 2023 Machinery Regulation is not just another compliance task – it is a blueprint for building resilient, intelligent machines.
“Can your FuSa & Cybersecurity teams prove – together – that the machine stays safe, even under attack or after the next update?”
Written by Antonio Tartaglione



