FAQ for developers and manufacturers on the EU AI Act

On May 21, 2024, the Council of the 27 EU Member States adopted the AI Act, a uniform framework for the use of artificial intelligence in the European Union, which has been in force since August 1, 2024. COMPAMED, the leading international trade fair for medical technology suppliers, answers the most important questions for developers of AI solutions in the medical technology industry.

The AI Act is the world's first comprehensive set of regulations for AI and is intended to create trust and acceptance in the technology. The AI regulation takes a risk-based approach and classifies AI systems into four categories according to their potential risks to society and individuals. The AI Act applies to all applications of AI. Medical devices are classified in the third category with a potentially high risk of harm, which is why far-reaching requirements are necessary.

The introduction of AI systems is seen as having major social benefits. They promise more economic growth and innovation as well as improved global competitiveness for the EU. The presumed risks, on the other hand, lead to legal uncertainty and may delay the use of AI. The AI Regulation is therefore intended to reduce the risks to health, safety and fundamental rights while improving the potential for the economy and society. AI real-world laboratories are to be made available free of charge to support European start-ups and SMEs in programming and developing compliant AI systems. They enable manufacturers and suppliers to develop and test AI systems independently under regulatory supervision.

Due to the classification of medical devices in the category with a potentially high risk of harm, a conformity assessment and a declaration of conformity must be drawn up for such systems. This classification includes all products that are assigned to risk class IIa or higher and MDR risk class I in the established Medical Device Regulation (MDR). In addition, further provisions of the In-Vitro Diagnostic Medical Devices Regulation (IVDR) must be complied with where applicable. This applies from the development process through the approval procedure to post-market safety.

Further requirements for software development include protection against cyber-attacks from outside. Any errors that occur in the software must be detected immediately, checked by humans and traced within the system.

Overall, the requirements are expected to lead to considerable delays and cost increases. In return, the AI Act is intended to guarantee a unique selling point for AI technology and thus a competitive advantage through EU-wide harmonization with the slogan "Made in Europe".

The AI Regulation will become fully applicable two years after its entry into force on August 2, 2026, but there are exceptions:

• As early as February 2, 2025, prohibitions, definitions and rules related to AI competencies will become applicable.
• On August 2, 2025, the provisions on governance and obligations relating to general purpose AI will become applicable.
• On August 2, 2027, the obligations for AI systems that are classified as high-risk AI systems because they are embedded in regulated products will become applicable.

By August 2025, each EU country must appoint a national AI supervisory authority. In addition to monitoring, this authority has the task of promoting innovation and competition. In Germany, this could be the Federal Network Agency, the data protection authorities or a new authority yet to be established. At EU level, the national supervisory authorities are represented in the Committee on Artificial Intelligence.

The penalties for violating the regulations are determined according to the size and business purpose of the company:

• For the use of prohibited AI systems: up to 35 million euros or 7 percent of the previous year's global turnover (whichever is higher).
• For breaches of obligations under the AI Act: up to 15 million euros or 3 percent of turnover.
• For providing false information: up to 7.5 million euros or 1.5 percent of turnover.

In May 2024, Federal Chancellor Scholz declared at an AI summit of heads of state and government that the Federal Government wants to make Germany a leading location for the development and application of AI technologies with a national AI strategy. To this end, it is supporting the research and application of AI, for example in newly established AI service centers for science and industry throughout Germany, especially for SMEs.

AI providers and software developers will gain access to larger markets thanks to greater legal certainty and uniform regulations. The regulations will only be applied where absolutely necessary and in such a way that companies are burdened as little as possible.

Real-world laboratories will be set up to enable testing under real conditions. Together with the AI centers of excellence and the public-private partnership for AI, data and robotics, as well as access to digital innovation centers and test and trial facilities, favorable framework conditions for the development and use of AI are to be created.