Navigating the EU AI Act and Machinery Regulation: A Guide for Companies

European companies developing and deploying artificial intelligence (AI) and robotics systems face a rapidly evolving regulatory landscape. The EU AI Act, coupled with the updated EU Machinery Regulation, introduces significant compliance requirements. Recent collaborative efforts, such as the “KIRR Real” project, are helping businesses in Baden-Württemberg, Germany, prepare for these changes, offering valuable insights for organizations across the European Union.

The Approaching Regulatory Deadlines

The stakes are rising as deadlines approach. Obligations for high-risk AI systems as a safety component will capture effect on August 2, 2027, marking the full application of the EU AI Act. Simultaneously, the recent EU Machinery Regulation will introduce new requirements for safety-relevant AI and self-learning systems starting in January 2027. These converging regulations demand proactive preparation from companies.

KIRR Real: A Real-World Laboratory for Compliance

Recognizing the demand for practical guidance, the Fraunhofer Institute for Production Technology and Automation IPA and ARENA2036 launched the “KIRR Real: Real Laboratory for Legally Compliant AI and Robotics” project in November 2024. The project, completed on December 31, 2025, involved collaboration with 15 companies to address the challenges of implementing the EU AI Act and the Machinery Regulation. The findings now serve as a resource for companies developing similar applications.

Understanding the Regulatory Approach

KIRR Real provided companies with “Legal Quick Checks” – concise consulting projects to assess their AI systems’ compliance. A legal firm classified each case into one of the four risk classes defined by the AI Act, ranging from minimal to unacceptable risk. Following this assessment, Fraunhofer IPA evaluated the technical documentation and underlying concepts of the AI development to determine legal conformity and provide actionable recommendations.

Key Challenges Identified

The project highlighted several key challenges companies face:

• Risk Classification: Determining the appropriate risk class for an AI system can be complex and dependent on specific employ cases, making resource planning difficult.
• Lack of Harmonized Standards: The absence of harmonized standards may necessitate third-party compliance testing for more AI systems.
• Technical Implementation: Companies grapple with questions regarding evidence provision, documentation requirements, and defining when AI becomes an integral part of a machine.
• Multiple Regulations: Navigating multiple applicable regulations and addressing concerns about unnecessary effort presents a significant hurdle.
• General-Purpose AI Models: Utilizing existing general-purpose AI models complicates the process of providing necessary evidence of compliance.

Integrating AI into Robotics and Machines

The new Machinery Regulation adds another layer of complexity, requiring companies to adapt existing products to the new rules within twelve months, in addition to complying with the AI Act’s requirements for AI in safety functions.

Recommendations for Action

Based on the KIRR Real project, several recommendations emerged:

• Practical Guidelines: There is a pressing need for concrete, easy-to-implement guidelines and structured procedures integrated early in the development process.
• Low-Threshold Support: Support mechanisms, such as real-world laboratories, should be accessible to AI development teams of all sizes.
• Early Engagement: Companies should seek support during early development phases, including legal assessments, workshops, and advice during system development.

Ongoing Efforts and Future Outlook

Fraunhofer IPA continues to analyze the project findings and contribute to the legislative process, actively participating in standardization committees and engaging with regulatory bodies like the Federal Network Agency and the Federal Institute for Occupational Safety and Health. The goal remains to refine AI regulation and provide optimal support to companies, ensuring responsible and beneficial AI innovation. Companies can contact Fraunhofer IPA to access these resources and collaborate on future initiatives.

The EU AI Act and the Machinery Regulation represent a significant shift in how AI and robotics are developed and deployed. Proactive preparation, coupled with access to practical guidance and support, will be crucial for companies to navigate this evolving landscape successfully.