Whether production companies, software providers or public administration-whoever uses AI systems are faced with new regulatory requirements. With the AI market surveillance law (Kimüg) Germany implements the EU Ki Ordinance (EU AI Act) nationally and creates clear responsibilities, control mechanisms and support offers. Therefore, the right time is now to deal with the new legal situation and to examine possible effects on existing or planned AI applications.
What is the Kimüg?
Table of Contents
Since August 1, 2024, the new AI regulation (KI-VO) has been in effect throughout the EU. It determines binding rules for how artificial intelligence developed, used and monitored, with the aim of protecting people and promoting trustworthy innovation. So that these rules are also implemented and controlled in each member state, all countries should be up to at the latest August 2025 create the necessary structures. In Germany, however, this procedure will be delayed. Due to the early Bundestag election in February 2025 and the associated transition phase, the legislative process cannot be locked before August 2, 2025. The Federal Ministry of Digital and State Modernization is now responsible.
National implementation in Germany takes over the so-called artificial intelligence market surveillance structure (Kimüg). It regulates which authorities will be responsible for monitoring the AI-VO in the future. The legislator consciously relies on existing structures to avoid duplication and use synergies. In addition to technical and legal control, the Kimüg also pursues the goal of cumulating the shortage of skilled workers in the AI area. Bundled responsibilities and support of other authorities are to be used efficiently and skills are to be built up efficiently.
In essence, it is about two things: the protection of fundamental rights, health and security and the promotion of trustworthy AI innovation. The Kimüg is the national implementation path for the EU AI Act, i.e. the Europe-wide AI regulation.
Which AI areas are affected by the Kimüg?
The EU Ki Ordinance introduces a risk-based approach: the greater the risk of fundamental rights, health or security, the more strictly. A distinction is made between four risk levels – from “inadmissible” to “low risk”. Two areas are particularly relevant for companies: forbidden AI practices and high-risk systems.
- High-risk-ki systems Are z. B. Applications used in security -critical areas: machine control, access to education or financial services, personnel decisions, critical infrastructures or health treatments. Strict requirements for data quality, transparency, risk management and human control apply here.
- Also AI systems with low risk Are not outside – providers will have to meet certain transparency obligations in the future, for example when users interact with a chat bot or a Deeppake. Important to know: Large AI models such as language models (GPT, Claude, etc.) do not fall under the national supervision, but are controlled directly by the new EU AI Office.
For companies, this means: Depending on which area a AI system is used, very different requirements can apply- from a complete ban to documentation and transparency obligations. A precise exam is crucial.
What does the Kimüg mean for companies?
The EU Ki Ordinance is calling for Member States to significantly strengthen AI competence in administration, industry and education. This means that specialists should become fit in dealing with AI-especially those responsible and developers who have to use and evaluate and evaluate AI systems safely and in accordance with the law. Regulated regulated industries such as industry, energy or public sector are particularly badly affected by the new rules. Here it is crucial to make AI solutions not only technically, but also legally secure. Small and medium -sized companies (SMEs) should also receive better access to further training, suitable tools and advice.
Many medium-sized companies, municipalities and public institutions use AI-based assistance systems, digital work instructions or automated evaluations to compensate for personnel bottlenecks. In accordance with the EU AI Act, the Kimüg creates important legal certainty here and can motivate companies and administrations to use AI solutions more specifically and safely.
What role does the Federal Network Agency play and why is that important?
For many companies, the Federal Network Agency will in future be a central contact for artificial intelligence. It takes over the market supervision wherever there has been no competent authority-for example, for general AI applications in everyday company life. Companies that use or develop AI systems should contact the Federal Network Agency at the latest if questions about classification, approval or complaint procedures arise. The Federal Network Agency should coordinate the implementation of the AI regulation in Germany, accept complaints, accompany developments And especially small and medium -sized companies (SMEs) help with new requirements. For this she builds several new structures:
- A central competence center (Kokivo) Supports companies-especially SMEs and start-ups-with questions about AI systems and their requirements. Such a competence center serves as a neutral point of contact, bundles knowledge and helps with the interpretation and practical implementation of regulatory requirements.
- One Market monitoring chamber Check particularly sensitive AI systems, for example in the areas of law enforcement or asylum procedures.
- When Central contact point the Federal Network Agency coordinates the exchange with the EU and fulfills reporting and reporting obligations.
- One Central complaint office It enables citizens to report possible violations of the AI regulation.
- In so -called KI-REALLABOREN Can companies test their systems under real conditions – with a special focus on practical innovation for medium -sized companies.
These structures are necessary in order to implement the requirements of the EU AI Act efficiently and in a practical way in Germany.
How does LMIS support companies in implementing the Kimüg?
As an experienced digitization partner, LMIS AG accompanies companies and authorities to implement AI solutions in accordance with the law, especially in regulated industries such as industry, energy or public sector. We know the requirements of the EU AI Act and support you compliance with legal requirements. Our solutions are designed to ensure both practical benefits and legal security. B. through transparent assistance systems, revision-proof data processing and modular AI components. SMEs in particular face the challenge of correctly classifying their AI systems, designing it in accordance with the law and reporting if necessary. This is where LMIS AG comes in: We develop AI solutions that have already thought of regulatory requirements-for example through comprehensible decision-making paths, logging functions or modular architectures that can be adapted to future requirements. LMIS AG ensures that AI does not become a compliance risk, but can be used strategically, legally and sustainably.
In a compact video recording, Prof. Dr. Marco Barenkamp, founder of LMIS AG, which requirements of the AI Act brings in practice and how companies can prepare for this at an early stage. After completing the form, the company sends you the recording by email.
Image of FreePik
