The EU-AI Act, a historic step in the regulation of artificial intelligence within the EU, sets strict requirements for AI usage. However, a comprehensive study by the Amsterdam University of Applied Sciences shows that many Dutch organizations are poorly prepared for this new legislation. With only a 56% average compliance score, the companies surveyed fall well below the required 100% compliance.

The study highlights the gap between large corporations and SMEs, with smaller enterprises less compliant with the requirements. Many organizations overestimate their level of compliance and lack a continuous monitoring system to ensure their AI systems do not evolve into unacceptable levels of bias. This is particularly problematic with systems such as recommendation algorithms that learn in real-time and can quickly lead to undesirable biases.

The study calls for urgent action from organizations to revise their approaches and invest in technologies that comply with the new regulations to foster innovation without crossing ethical boundaries. This requires a strategic reevaluation of how AI is deployed and managed, especially with the threat of outsourcing and offshoring that can undermine compliance.