| Provisions of the AI Act | Realization period |
| AI literacy: Art. 4 AI Act | February 2, 2025 |
| Prohibited AI Systems: Art. 5 AI Act | February 2, 2025 |
| Obligations for GPAI models: Art. 51 ff. | August 2, 2025 |
| General deadline for implementing the AI Act | August 2, 2026 |
| Transparency obligations: Art. 50 AI Act | August 2, 2026 |
| Obligations for high-risk AI systems according to Art. 6 para. 2 in conjunction with Annex III AI Act |
August 2, 2026 |
| Obligations for high-risk AI systems according to Art. 6 para. 1 in conjunction with Annex I AI Act |
August 2, 2027 |
Implementation of the AI Act
The implementation of the AI Act has many peculiarities:
- Legal uncertainty: Due to the still unclear wording and provisions of the AI Act, there is a degree of uncertainty as to how certain provisions should be interpreted and implemented in practice. This requires operational decisions to ensure consistent interpretation and implementation across the organization. Such an approach can help minimize risk and ensure consistent compliance.
- Escaping the offense: The AI Act provides some structuring options that may allow companies to escape the scope of the Act or to rely on certain exemptions (e.g. for research and development). Companies would be well advised to use these structuring options and design AI systems in a way that avoids falling under the strict provisions of the AI Act from the outset.
- Enforcement: There are still numerous question marks surrounding the public enforcement of the AI Act. The competent authorities face the challenge of recruiting a sufficient number of qualified personnel capable of interpreting the legally and technically complex requirements of the AI Act in a practical manner and enforcing them in a targeted manner. Irrespective of this, there is a risk of private enforcement. The AI Act, as a protective law for the benefit of third parties (Section 823 (2) of the German Civil Code), offers numerous points of contact for private plaintiffs, especially since a lack of AI competence (Section 4 of the AI Act) could constitute grounds for an accusation of negligence. To protect themselves from potential lawsuits, companies should take a variety of organizational measures and establish a compliance structure.
- Take stock: A comprehensive professional assessment of the current situation is a challenge for many companies. To be as efficient as possible, a standardized questionnaire is recommended.
- Gantt-Chart: The implementation of the AI Act involves many feedback loops and requires operational fine-tuning that takes into account the temporal complexity and allows for safety buffers..
Advice at the highest level
The prohibitions of the AI Act pose major challenges for the development and operation of AI. Due to the short implementation period, there is already an enormous need for action today, for example in the implementation of new products and business models. Companies would be well advised to take a close look at the requirements.
The implementation of the AI Act often cannot be forced into templates, especially when it comes to new AI tools. Our consulting approach is to first gain a technical understanding of the planned technology and its opportunities and challenges for your business model. We then work with you to develop optimal strategies and solutions.
Aitava is a leading AI law firm. We provide continuous and comprehensive advice on AI law and provide decisive impulses. Pragmatic and to the point.
