When the EU AI Act entered into force on August 1, 2024, many companies treated it as a future compliance topic.
In 2026, it is no longer theoretical.
Belgian businesses are now entering the phase where transparency rules, high-risk obligations, and governance requirements become operational realities. The question is no longer whether AI will be regulated. The question is whether organizations are structurally ready.
And for many sectors, especially HR, recruitment, and financial services, the impact is deeper than expected.
The 2026 turning point: where we stand now
The implementation of the EU AI Act is progressive, but 2026 marks a major inflection point.
Already in force:
- February 2025: Ban on unacceptable-risk systems + mandatory AI literacy
- August 2025: Rules for General Purpose AI models (GPAI)
Coming into full effect:
- August 2026: Broad applicability, including high-risk systems
- August 2027: High-risk AI embedded in regulated products
This means Belgian companies must now actively:
- Classify their AI systems by risk level
- Implement governance and documentation
- Ensure transparency toward users
- Establish human oversight mechanisms
Non-compliance is not symbolic. Fines can reach up to €35 million or 7% of global turnover.
What changes operationally for Belgian companies?
The AI Act does not only affect AI developers. It applies equally to “deployers” companies that simply use AI tools in their daily operations.
Three immediate impact areas stand out in 2026:
1. AI Literacy (since February 2025)
Companies that provide or deploy AI systems are legally required to take measures to ensure that staff using or operating these systems have a sufficient level of AI literacy – meaning they understand how the tools work, their limitations, risks and potential impacts.
This is not optional training. It is a compliance obligation.
2. Transparency (August 2026)
Organisations must ensure appropriate AI transparency, including clearly informing users when they interact with AI‑powered chatbots, AI‑generated or manipulated content, or automated decision‑making systems. This applies internally (employees) and externally (clients).
3. GPAI responsibility (since August 2025)
If your company integrates general‑purpose AI models into services or workflows, you need to ensure that your provider complies with EU transparency, copyright and safety requirements – and that your own use of the model stays within those rules. The responsibility does not disappear because the model is external.
Why HR and Recruitment Are Directly Exposed
Under Annex III of the AI Act, most AI systems used in HR are classified as high-risk.
This includes:
- CV screening software
- AI-based candidate ranking tools
- Performance evaluation algorithms
- Promotion recommendation systems
- Employee monitoring technologies
Why such strict classification?
Because these systems directly influence access to employment and career progression.
From August 2026, high-risk HR AI requires:
- Full transparency toward candidates and employees
- Clear explanation of how decisions are influenced by AI
- Demonstrable bias mitigation
- Representative training data
- Continuous human supervision
- Automatic logging for traceability
Candidates will have the right to understand how an AI system influenced their rejection.
This fundamentally reshapes recruitment governance.
Financial Services: AI Under Heavy Scrutiny
Banks and insurers are equally exposed.
High-risk systems include:
- Credit scoring
- Loan approval algorithms
- Fraud detection models
- Risk pricing systems
From 2026 onward, institutions must:
- Document model functioning in detail
- Conduct fundamental rights impact assessments
- Guarantee absence of discriminatory bias
- Retain automated logs for each decision
In practical terms, this means compliance, IT, and risk departments must collaborate far more closely than before.
AI becomes a governance matter, not only a performance tool.
Other High-Risk Sectors You Should Not Ignore
Beyond HR and finance, the AI Act identifies several additional high-risk domains:
- Biometric identification and emotion recognition
- Critical infrastructure management (energy grids, traffic systems)
- Education and exam evaluation
- Law enforcement and judicial support
- Migration and border control
- Access to public services and social benefits
- Healthcare diagnostics and medical AI
Each requires:
- Lifecycle risk management
- Human oversight
- High-quality datasets
- Traceable logging systems
This extends the AI Act’s impact far beyond tech companies.
What This Means for SMEs and Start-ups
Belgium has introduced regulatory sandboxes allowing companies to test AI solutions in collaboration with regulators such as the IBPT.
Sanctions remain proportionate for SMEs in cases of non-intentional non-compliance.
However, proportionality does not mean exemption. Governance structures are still required.
The Governance Shift Companies Underestimate
The AI Act forces organizations to answer uncomfortable questions:
- Do we have a complete inventory of AI tools used internally?
- Who is accountable for AI decisions?
- Can we explain how our models work?
- Are we documenting our systems properly?
- Is there a formal AI policy in place?
Belgian experts now strongly recommend establishing an internal AI Policy defining:
- Acceptable use
- Risk tolerance
- Responsibility allocation
- Human oversight protocols
AI is no longer an isolated innovation topic. It becomes a board-level governance issue.
A Practical Starting Point
Companies unsure about their exposure can begin with the official EU AI Act Compliance Checker.
This tool helps determine whether a specific AI system may fall under high-risk classification.
It is not a substitute for legal advice, but it is a valuable first diagnostic step.
Belgium’s Position: A Risk or an Advantage?
Belgium is not lagging behind.
Approximately 20% of Belgian companies already use AI, above the EU average. National and regional initiatives such as the National Convergence Plan for AI, DigitalWallonia4.ai, the Flemish AI Plan, and FARI in Brussels reinforce a strong innovation ecosystem.
The AI Act does not aim to slow innovation. It aims to structure it.
Organizations that integrate compliance early may gain a strategic advantage in trust, reputation, and governance maturity.
The Real Question for 2026
The EU AI Act is not just a regulatory shift.
It is:
- A governance shift
- A transparency shift
- A talent shift
- A cultural shift
Companies that see it purely as a legal burden will struggle. Companies that treat it as an organizational upgrade will be better positioned for long-term digital resilience.
The transition phase is over and the operational phase has begun.
Resources.avif)
