AI Governance: Beyond Compliance to Ethics

Three Tiles of Wisdom: Lessons for AI Governance from the Dutch Experience

Joost Gerritsen, CEO of Digibeetle, lawyer and affiliate researcher at Utrecht University, recently presented to the Vlaamse Toezichtcommissie about the critical intersection of AI, algorithms, and human rights. Following Dutch tradition, he brought three virtual tiles – each inscribed with crucial wisdom for anyone working in AI Act compliance and data protection.

These lessons, drawn from real-world algorithmic failures and regulatory responses, illuminate why supervisory authorities, privacy professionals, and legal consultants must look beyond minimum compliance toward genuine human rights protection.

First Tile: The Human Heart Behind Every Data Point

Data may be the oil of AI, but beneath every number beats a human heart, Joost reminded the commission. The Dutch childcare benefits scandal serves as a devastating case study of what happens when we forget this fundamental truth.

The numbers tell a story of systemic failure:

• 70,000 children affected by algorithmic discrimination
• 2,090 children removed from their homes
• €30,000 median damage per family
• €2.75 million fine from the Dutch Data Protection Authority for discriminatory profiling

Behind every data point was a family whose life was destroyed by an algorithm that saw risk profiles instead of people. The system flagged dual nationality as a risk factor, turning a characteristic protected under fundamental rights law into a weapon of discrimination. This wasn’t a technical failure – it was a human one, where efficiency metrics overshadowed human dignity.

For privacy professionals and supervisory authorities, this scandal underscores why GDPR enforcement must centre on human impact, not just technical compliance. Every algorithmic decision affects real lives, real families, real futures.

Second Tile: Smart Machines Require Wise Humans

The second lesson highlights a critical governance failure: Smart machines require wise humans to guide them. Sandra Palmen, a senior civil servant, warned about the childcare benefits system’s failures in 2017. Her prescient memo was buried, and she was sidelined – a cautionary tale about organisational cultures that silence dissent.

This is precisely why tools like IAMA (Impact Assessment for Human Rights and Algorithms) and the AI Performance Review have become essential for EU digital compliance. These frameworks force organisations to confront uncomfortable questions:

• Which fundamental rights are at stake?
• Is the algorithmic impact proportionate to the goal?
• Can we explain our decisions to affected individuals?
• Do we have meaningful human oversight mechanisms?

Smart AI demands not just technical expertise but wise governance, courageous whistleblowers, and leaders who listen. The technology in the Dutch scandal worked exactly as designed – but the design itself violated fundamental principles of fairness and non-discrimination. Algorithmic accountability requires creating organisational cultures where ethical concerns can be raised without fear of retaliation.

Third Tile: The AI Act Sets the Floor, Not the Ceiling

The final wisdom challenges the compliance mindset: The AI Act sets the floor, not the ceiling – I expect us to reach higher. While the AI Act establishes important standards, treating it as the endpoint of ethical AI governance would be a grave mistake.

Real protection requires going beyond minimum compliance:

• Explainability: Algorithms must be understandable not just to engineers but to affected individuals
• Bias mitigation: Datasets must be continuously monitored for discriminatory patterns
• Scientific rigour: Methods must meet peer-review standards, not just regulatory checkboxes
• Human intervention: Meaningful ability to override algorithmic decisions must exist
• Transparency: Regular public reporting on algorithmic impacts and corrections

The Act’s numerous exceptions and provisos already reveal its limitations. As legal consultants and privacy officers, we must recognise that laws provide the minimum baseline while ethics demand more. Organisations serious about fundamental rights protection will develop governance frameworks that exceed regulatory requirements.

From Lessons to Action: Implementing Ethical AI Governance

These three tiles of wisdom translate into concrete actions for organisations deploying AI systems:

1. Centre human impact in all algorithmic assessments, moving beyond technical metrics to understand real-world consequences
2. Create safe channels for internal dissent and ethical concerns about AI systems
3. Develop governance frameworks that exceed AI Act minimums, incorporating best practices from supervisory authority guidance
4. Implement continuous monitoring for discriminatory patterns and unintended consequences
5. Maintain genuine human oversight with power to intervene and correct algorithmic decisions

The Role of Legal Intelligence in Ethical AI Governance

As the Dutch childcare scandal demonstrates, preventing algorithmic harm requires more than good intentions. It demands comprehensive understanding of how European courts and supervisory authorities interpret fundamental rights in algorithmic contexts. Legal professionals need access to:

• Court decisions on algorithmic discrimination and profiling
• Supervisory authority enforcement actions and fine calculations
• Cross-jurisdictional approaches to AI governance
• Evolving interpretations of proportionality and necessity in automated decision-making

Staying Ahead of AI Governance Challenges

The lessons from Joost’s presentation to the Vlaamse Toezichtcommissie underscore a critical reality: effective AI governance requires continuous learning from both failures and successes across Europe. As AI Act implementation unfolds, tracking how different member states and supervisory authorities approach algorithmic accountability becomes essential.

This is where Digibeetle’s academic heritage proves invaluable. As a lawyer and affiliate researcher at Utrecht University, Joost understands the challenges of accessing comprehensive, current legal intelligence for both research and practice. Our expert-curated legal database ensures you have immediate access to:

• Court rulings on algorithmic discrimination like the Dutch childcare scandal
• Supervisory authority decisions on AI and automated decision-making
• Cross-referenced fundamental rights assessments and impact studies
• Daily updates on AI governance enforcement across Europe

Beyond our comprehensive database, we support your AI literacy journey through expert-led education. Start with our free webinar on safeguarding fundamental rights under the AI Act, which provides essential context for understanding why governance must extend beyond technical compliance – exactly the lessons highlighted in these three tiles of wisdom.

Whether you’re developing AI governance frameworks, advising on GDPR and AI Act compliance, or researching algorithmic accountability, Digibeetle transforms scattered legal intelligence into actionable insights. We help you learn from Europe’s collective experience to prevent tomorrow’s algorithmic disasters.

Ready to elevate your AI governance beyond compliance? Start your 30-day free trial to access comprehensive legal intelligence on algorithmic accountability and fundamental rights protection. Or book a consultation to discover how we support legal professionals and researchers in navigating the complex intersection of AI, law, and human rights.