Detailed overview
Belgium is regulated by the EU AI Act. Belgian companies and public authorities using, developing, importing or distributing AI systems must classify the AI system under the EU AI Act and apply the relevant duties. Prohibited AI is banned, high-risk AI is subject to strict controls, certain AI systems must meet transparency duties, and general-purpose AI models have documentation, transparency and copyright-policy obligations.
High-risk AI includes AI used in areas such as recruitment, worker management, education, access to essential services, credit scoring, insurance risk assessment, biometric identification, critical infrastructure, law enforcement, migration, border control and justice. Providers of high-risk AI must maintain a risk-management system, data-governance controls, technical documentation, records, human oversight, accuracy, robustness and cybersecurity. Deployers must use the system according to instructions, monitor it and inform affected persons where required.
Belgium's national AI policy is based on the National Convergence Plan for the Development of Artificial Intelligence, approved by the Council of Ministers on 28 October 2022. The plan sets nine objectives: promoting trustworthy AI, guaranteeing cybersecurity, increasing competitiveness and attractiveness through AI, developing a data-driven economy and infrastructure, placing AI in healthcare, supporting sustainable mobility, protecting the environment, improving lifelong training, and providing citizens with better services and protection.
Belgium's federal administration also provides for a steering committee jointly set up by FPS BOSA and FPS Economy. This committee supports implementation of the national AI plan, makes proposals on the AI programme and acts as a central federal contact point for AI questions.
AI involving personal data is also regulated under GDPR. The Belgian Data Protection Authority has launched an "AI & Data Protection" initiative and published materials explaining how AI systems affect privacy, what personal data they process, and what rights individuals have under data-protection law.
Penalties depend on the breached framework. EU AI Act fines apply to AI Act breaches. GDPR penalties apply where AI involves unlawful personal-data processing. Sectoral penalties may apply in finance, healthcare, telecoms, employment, consumer protection or public administration. Under the EU AI Act, prohibited AI breaches may reach EUR 35 million or 7% of worldwide annual turnover, while many other operator breaches may reach EUR 15 million or 3%.
Practical requirements & details
Sourced from Regulation (EU) 2024/1689 (the EU AI Act) and Belgium's National Convergence Plan for the Development of Artificial Intelligence (approved 28 October 2022).
EU AI Act classifications
- Prohibited AI — banned.
- High-risk AI — strict controls.
- Transparency-risk AI — disclosure duties.
- GPAI models — documentation, transparency and copyright-policy obligations.
High-risk areas
- Recruitment, worker management, education, access to essential services, credit scoring, insurance risk assessment, biometric identification, critical infrastructure, law enforcement, migration, border control and justice.
Provider duties
- Risk-management system, data-governance controls, technical documentation, records, human oversight, accuracy, robustness and cybersecurity.
Deployer duties
- Use the system according to instructions, monitor it and inform affected persons where required.
National Convergence Plan — nine objectives
- Promoting trustworthy AI; guaranteeing cybersecurity; increasing competitiveness; developing a data-driven economy; placing AI in healthcare; supporting sustainable mobility; protecting the environment; improving lifelong training; providing citizens with better services and protection.
FPS BOSA + FPS Economy steering committee
- Joint federal committee supporting national AI plan implementation.
- Acts as a central federal contact point for AI questions.
Data-protection overlay
- Belgian Data Protection Authority's AI & Data Protection initiative explains how AI affects privacy and data-subject rights.
Penalties
- EUR 35 million or 7% of worldwide annual turnover — breaches of prohibited AI rules.
- EUR 15 million or 3% of worldwide annual turnover — breaches of many other AI Act operator obligations.
- EUR 7.5 million or 1% of worldwide annual turnover — supplying incorrect, incomplete or misleading information to authorities.
Related entries
See also the European Union entry, which covers the EU AI Act (Regulation (EU) 2024/1689) — the substantive framework that this jurisdiction implements and supervises domestically.