Artificial Intelligence Act (AI Act)

Basics and objectives

The Artificial Intelligence Act comes into force on August 1, 2024 as the world’s first comprehensive AI regulation. The regulation creates a uniform legal framework for the development and use of AI systems in the European Union. The law follows a risk-based approach and categorizes AI applications according to their risk potential. The regulations aim to protect fundamental rights, democracy and the rule of law. The regulation prohibits certain AI applications that endanger civil rights. It will be implemented in stages, starting with the bans from February 2025, with full application planned from August 2026. The regulation applies to all companies that offer or use AI systems in the EU. The regulations also apply to providers outside the EU whose AI systems are used in the EU. Innovation is to be promoted through clear framework conditions. Europe’s leading role in the field of ethical AI will be strengthened. The competitiveness of Europe as a business location will be secured.

Risk-based regulatory approach

The AI Act distinguishes between four risk categories for AI systems with different requirements. Unacceptable risks lead to a complete ban on certain AI applications. High risks require comprehensive obligations and conformity assessments. Limited risks are subject to transparency obligations towards users. Minimal risks remain largely unregulated. Classification is based on clear criteria and use cases. The categorization is regularly reviewed and adjusted. The proportionality of the requirements is ensured. Innovation is enabled by differentiated regulations. Legal certainty is created through clear categories. The protective effect is concentrated on critical applications. Flexibility for non-critical applications is maintained.

Prohibited AI practices

The regulation explicitly prohibits AI systems that jeopardize fundamental rights. Biometric categorization according to sensitive characteristics is prohibited. The creation of facial databases through untargeted scraping is prohibited. Emotion recognition in the workplace and in schools is not permitted. Social scoring and behavior-manipulative AI systems are prohibited. Predictive policing based purely on personal profiles is prohibited. Manipulation of human behavior by AI is prohibited. The exploitation of vulnerabilities is prevented. Fundamental rights are comprehensively protected. Human dignity is safeguarded. Social stability is ensured. Democratic values are defended.

Requirements for high-risk AI

High-risk AI systems must meet extensive requirements and undergo conformity assessments. The systems must be transparent and comprehensible. The data sets must be of high quality and non-discriminatory. Human supervision must be guaranteed at all times. Documentation must be complete and up to date. Risk management systems must be implemented. Cybersecurity must be guaranteed at a high level. Conformity assessment must be carried out before market launch. Continuous monitoring must be ensured. Liability must be clearly regulated. Responsibilities must be defined. Traceability must be ensured.

Supervision and enforcement

Enforcement is carried out by national supervisory authorities in coordination with the European Artificial Intelligence Board. Sanctions can amount to up to 35 million euros or 7% of annual global turnover. The authorities will be given extensive powers of investigation and enforcement. Cross-border cooperation will be intensified. Market surveillance will be carried out systematically. Conformity assessments will be strictly controlled. Complaints mechanisms will be established. Transparency is increased through reporting obligations. Effectiveness is regularly evaluated. Sanctions have a preventive effect. Enforcement is standardized throughout the EU. International cooperation is strengthened.

Practical implementation

The implementation of the AI Act requires companies to make extensive preparations. The risk analysis of AI systems must be carried out systematically. Compliance structures must be set up. The technical systems must be adapted. All documentation must be created. Employees must be trained. Processes must be defined. Responsibilities must be assigned. Controls must be implemented. Reporting channels must be established. Emergency plans must be drawn up. Resources must be made available.

Future prospects

The AI Act will have a lasting impact on the development of artificial intelligence in Europe. Ethical AI development will become a locational advantage. The international role model function will be strengthened. Legal certainty promotes investment. Innovation is supported by clear rules. Competitiveness is secured in the long term. Social acceptance is increased. Technological development is managed responsibly. European values are safeguarded. Global standard setting is influenced. Cooperation is intensified. Future viability is secured.