One regulation, not two: what the EU AI Act Omnibus means for machinery manufacturers

10 Sep 2026

The EU AI Act Omnibus streamlines compliance for AI-enabled machinery, shifting requirements into the Machinery Regulation and simplifying CE marking.

On 29 June 2026, the Council of the EU gave its final green light to the Digital Omnibus on AI - a key deliverable in the EU’s wider simplification agenda - completing the legislative process after the European Parliament's approval on 16 June, 2026. The regulation will now be published in the Official Journal and enter into force shortly after. For most observers, the headline was the delayed application of high-risk AI rules. For machinery manufacturers placing AI-enabled products on the EU market, the more consequential news sat further down the press release: the Machinery Regulation (EU 2023/1230) is now exempt from direct applicability of the AI Act.

What actually changed for the machinery sector

Until this agreement, manufacturers of machinery containing AI as a safety-relevant component faced parallel compliance under two horizontal regimes (the AI Act and the Machinery Regulation). Each came with its own conformity assessment logic, its own technical documentation expectations, and its own set of harmonized standards (many of which are still in development).

The Omnibus collapses these requirements into a single route. The Machinery Regulation becomes the channel through which AI-specific obligations reach the product. The Commission has been empowered to adopt delegated acts under the Machinery Regulation that introduce health and safety requirements for AI systems classified as high-risk under the AI Act. The substance of AI Act Articles 9–15 is expected to flow into the Machinery Regulation through those delegated acts. These requirements include: 

• Risk Management 

• Data Governance 

• Technical Documentation 

• Logging 

• Transparency 

• Human Oversight 

• Accuracy 

• Robustness, and 

• Cybersecurity

Does the AI Act apply to my machine?

No, it won’t. Manufacturers will not need to perform a separate AI Act conformity assessment, and will not be relying on AI Act harmonized standards as the primary route to presumption of conformity. The CE marking pathway runs through the Machinery Regulation, just as it does today, but the essential health and safety requirements the equipment must meet will expand to cover the AI components inside the product.

Do high-risk requirements still apply?

In substance, yes. The delegated acts will add health and safety requirements in respect of AI systems that are classified as high-risk under the AI Act. Manufacturers should plan for a compliance burden materially equivalent to the AI Act high-risk regime — risk management systems, data quality controls, traceability and logging, human oversight provisions, robustness and accuracy validation — but evidenced through Machinery Regulation technical documentation and notified body engagement where applicable.

The new application date for high-risk AI systems embedded in products is 2 August 2028.

Which standards matter

The standards landscape has not been redrawn, but it has been clarified. Three layers are most relevant:

Machinery functional safety.

ISO 13849-1, ISO 13849-2, IEC 62061 and IEC 61508 remain the foundation. Your safety architecture continues to be assessed against these. 

AI and functional safety.

ISO/IEC TR 5469 — the technical report on functional safety and AI — (soon to be superseded by its successor ISO/IEC TS 22440 currently in development) is the most directly relevant document for safety-relevant AI in machinery today. Expect this and its successors to do significant work in bridging traditional safety engineering with AI-specific concerns. IEC 61508 is currently under revision to explicitly make the link to these AI standards.

AI management and lifecycle.

Supporting standards like ISO/IEC 42001 (AI management systems), ISO/IEC 23894 (AI risk management), ISO/IEC 5338 (AI system lifecycle) and ISO/IEC 22989 (concepts and terminology) provide the organizational and process scaffolding around the engineering work.

The open question is whether the harmonized standards being developed by CEN/CENELEC JTC 21 under the AI Act standardization request will ultimately be cited in the Official Journal under the Machinery Regulation, giving presumption of conformity through the new delegated acts. Practical alignment would be a sensible outcome, but the answer depends on the standardization request that follows the delegated acts and how the Commission frames it.

Don’t wait to align your designs with these requirements, if you wait it may result in costly delays. The JTC 21 standards in development for the AI Act (prEN 18228 risk management, prEN 18284 data quality and governance, prEN 18229 parts 1–5 AI trustworthiness framework, prEN 18282 Cybersecurity and prEN 18286 quality management system) are not yet harmonized, but they encode exactly the engineering substance you will be assessed against.

Manufacturers who maintain a traceability matrix from Machinery Regulation EHSRs through AI Act Arts. 9–15 to these standards will be in a strong position both for CE conformity today and for the delegated-act regime in 2028.

What this means for your roadmap

The Machinery Regulation conformity route will have a single regulatory anchor for AI-enabled machinery — build your compliance program around it. The substance of the AI Act high-risk regime should be assumed to apply, even though formal applicability runs through the Machinery Regulation. The 2 August 2028 date is the target for AI-specific obligations on embedded high-risk systems; the runway is real, but not generous when standards, delegated acts and conformity infrastructure are still being built. Watch for the Commission’s promised guidance on minimizing compliance burden where sectoral and AI Act requirements interact. And treat the Omnibus as direction-of-travel, and watch for the delegated acts themselves and the standards citations that are still to come.

How Intertek can support 

Intertek can support manufacturers by helping translate these regulatory changes into practical compliance actions. 

Through Intertek AI², our independent AI assurance program, we help organizations assess whether AI-enabled products and systems are smart, safe, secure, and trusted. For machinery and robotics manufacturers, that means looking beyond whether the AI works and focusing on whether it can be safely integrated into the machine’s overall risk profile. Intertek AI² supports areas such as AI risk assessment, performance reliability, safety, data governance, cybersecurity, and lifecycle controls. 

In practice, this can include reviewing risk assessments, identifying where AI functions may influence safety, assessing technical documentation, and supporting conformity planning under the Machinery Regulation. For robotics and autonomous machinery, it also means considering functional safety, human–machine interaction, cybersecurity, and how software or AI updates may affect continued compliance. 

The objective is not just to prepare for a deadline. It is to build a defensible safety case for AI-enabled machinery.