On August 1, 2024, the European Union's AI Act came into force – the world's first comprehensive law to regulate artificial intelligence. But what does this mean specifically for businesses, developers, and users? In this article, we provide an overview of the most important regulations and their implications.
What is the AI Act?
The AI Act is an EU regulation that governs the use of artificial intelligence in Europe. Its goal is to protect citizens' fundamental rights, promote innovation, and at the same time create clear rules for the development and use of AI systems.
The Risk-Based Approach
The AI Act follows a risk-based approach. AI systems are classified into four categories according to their risk potential:
1. Prohibited AI Applications
Certain AI applications are fundamentally prohibited in the EU because they are considered too risky:
- Social Scoring – Rating systems that classify people based on their behavior (as in China)
- Real-time biometric surveillance in public spaces (with few exceptions for law enforcement) – from February 2025
- Manipulation of human behavior – AI systems that exploit vulnerabilities of individuals
- Indiscriminate collection of biometric data
2. High-Risk AI Systems
AI systems used in sensitive areas are considered high-risk and are subject to strict requirements:
- Critical infrastructure (e.g., water supply, transport)
- Education and vocational training
- Employment and human resources
- Law enforcement and justice
- Medical devices
For these systems, the following applies from August 2026:
- Risk assessments and management
- High data quality and transparency
- Human oversight
- Comprehensive documentation requirements
- Conformity assessments by third parties
3. AI Systems with Minimal Risk
Most AI applications fall into this category – such as spam filters, AI-powered video games, or recommendation systems. No special requirements apply to them.
4. General Purpose AI (GPAI)
For large AI models like GPT-4, Claude, or Gemini, special regulations apply from August 2025:
- Transparency obligations (e.g., disclosure of training data)
- Special requirements for models with systemic risks
- Compliance with copyright regulations
Who is Affected?
The AI Act affects:
- AI developers and providers – They bear the main responsibility for compliance with the regulations
- Companies using AI – Users of AI systems must also ensure that the systems used comply with the regulations
- Importers and distributors – Anyone who imports or distributes AI systems into the EU must ensure their conformity
Enforcement and Penalties
The AI Act is enforced by national authorities in EU member states. Violations can result in hefty fines:
- Up to 35 million euros or 7% of global annual turnover (whichever is higher) for violations of prohibited AI practices
- Up to 15 million euros or 3% of turnover for violations of other provisions
Timeline: When Do the Regulations Take Effect?
- February 2025 – Prohibition of certain AI practices (e.g., real-time biometric surveillance)
- August 2025 – Regulations for General Purpose AI (GPAI)
- August 2026 – Full application of provisions for high-risk AI systems
What Does This Mean for Developers and Businesses?
The AI Act brings new responsibilities – but also opportunities:
- Prepare early – Companies should already review their AI systems and adapt them to the new requirements
- Create transparency – Documentation and traceability become central
- Build trust – Compliance with the AI Act can become a competitive advantage
Conclusion
The AI Act is a milestone in the regulation of artificial intelligence. It creates clear rules, protects fundamental rights, and at the same time promotes innovation. For businesses, this means: Act now, review systems, and ensure compliance.
The coming years will show how the regulation proves itself in practice – and what impact it will have on the AI landscape in Europe and worldwide.