EU 'Digital Omnibus' creates safety gap for AI-powered machinery
A regulatory shift removing machinery from the AI Act's high-risk regime leaves worker safety protections in limbo until 2028.
The European Union has created a significant regulatory void for industrial safety by removing AI-powered machinery from the AI Act's high-risk regime. This shift, enacted through the 'Digital Omnibus on AI' in May 2026, decouples critical safety components from strict AI oversight to streamline compliance for manufacturers.
Under the new rules, machinery products and their AI-driven safety systems are no longer subject to the AI Act's directly applicable high-risk requirements. Instead, the European Commission has until August 2028 to transfer necessary AI-specific safeguards into the existing Machinery Regulation via a delegated act. While the move aims to boost competitiveness by avoiding duplicate regulations, it leaves a gap in legally binding protections for several years.
The Regulatory Gap
The AI Act was designed to impose rigorous obligations on high-risk systems, specifically targeting data governance, bias mitigation, cybersecurity, and quality management. These specific obligations are not currently present in the Machinery Regulation. Consequently, AI systems used for critical functions—such as collision avoidance or emergency braking in forklifts—may operate without the same level of legal scrutiny regarding their underlying data or algorithmic bias.
While manufacturers can temporarily rely on AI Act standards to guide their development, these standards lack the force of law. The European Trade Union Confederation (ETUC) warned that standards cannot replace the law, noting that while they explain how to meet requirements, they cannot create legal protections that the legislation itself does not guarantee.
Implications for Worker Safety
The decoupling of these regulations creates what critics describe as a 'regulatory lay-by.' If AI-powered safety components are not subject to robust, legally binding safeguards, workers face increased risks of injury or death from system failures. Because the AI Act's requirements for transparency and quality management are missing from the current Machinery Regulation, failures in these systems may be poorly tested or entirely untraceable.
This transition period introduces substantial legal uncertainty. By lowering the immediate administrative burden on manufacturers, the EU may inadvertently allow sub-standard safety systems to enter the market before the 2028 deadline.
What to Watch
The industry now awaits the European Commission's delegated act, which must reconcile the AI Act's technical safeguards with the Machinery Regulation. The ETUC has emphasized that while the regulatory route may change, the destination must remain the highest possible level of safety for workers. Until August 2028, the primary concern remains whether voluntary standards will be sufficient to prevent workplace accidents in an increasingly automated industrial landscape.