Berlin, May 7, 2026 – The TÜV Association sees both positives and negatives in the agreement reached during the trilogue negotiations on the “AI Omnibus.” Positive aspects include adjustments such as longer implementation deadlines for the AI Act, reporting requirements for AI security incidents, and a ban on sexualized deepfakes and so-called “nudifier” applications.
The TÜV Association is critical of the fact that machines fall outside the direct scope of the AI Act and that safety requirements are instead to be regulated in the subordinate Machinery Regulation. The goal of the AI Act is to minimize the risks of high-risk AI systems in various product groups such as medical devices, machines, and even toys. “The ‘sector exit’ for machines leads to regulatory fragmentation with differing requirements, procedures, and deadlines. The result is legal uncertainty and more—rather than less—bureaucratic burden for companies,” says Dr. Joachim Bühler, Managing Director of the TÜV Association. “The AI Omnibus has done a disservice to efforts to reduce bureaucracy. Now it is crucial to ensure an equally high level of protection in the Machinery Regulation.” This includes, among other things, requirements regarding risk and quality management, human oversight, the robustness of AI systems, and the reporting of serious safety incidents.
From an industrial policy perspective, too, the sector-specific approach is not effective. “In Europe, industrial AI standards for the global market are currently being developed based on the AI Act,” says Bühler. “The shift to the Machinery Regulation threatens to further delay the development of sector-specific AI standards because the legal basis is currently lacking. Europe is currently squandering the opportunity to take a global leading role in AI standards.”
The TÜV Association warns against excluding further product groups, such as medical devices, from the AI Act until it is implemented. Otherwise, there is a risk of a patchwork of individual regulations across numerous sector-specific legal acts. The TÜV Association has serious concerns about the omnibus procedure and the trilogue negotiations. “The AI Act, with its horizontal structure and risk-based approach, is a piece of legislation that has been developed over many years with the active involvement of a wide range of stakeholders,” says Bühler. “It is unacceptable for the fundamental architecture of the law to be dismantled in a cloak-and-dagger operation. Such fundamental changes require transparency, thoroughness, and public debate.” In doing so, politicians are undermining trust in the European legislative process.
