The Dutch Data Protection Authority AP (Autoriteit Persoonsgegevens) has called on the new cabinet to urgently clarify rules for artificial intelligence. The privacy supervisor warned that it is currently unable to tackle unsafe or discriminatory algorithms and is pushing for stronger oversight powers.
Nine indicator tracking
The authority uses a “barometer” comprising nine indicators to track the impact of AI. Six months ago, the authority marked two indicators red: AI systems lacked sufficient registration, and officials lacked insight into incidents. Since then, two more indicators have turned red. The watchdog says the government has not yet properly regulated the legal framework for supervision, and AI systems lack clear standards.
Concerns
The AP also raised concerns about the protection of young people. Many children use AI not only for schoolwork but also as digital friends to chat with. The watchdog warns that young people may become addicted to these tools and lack the ability to properly assess the risks involved. The authority identified the most significant dangers as the rise of deepfakes, AI-driven fraud, and psychological harm caused by chatbots. It also noted that security measures are failing to keep pace with rapid technological developments.
According to the AP, some organisations appear to be avoiding their responsibilities. High-risk AI systems used in healthcare and crime detection must be registered, but some groups are reportedly labelling them as “ordinary algorithms” to bypass stricter rules. Starting next year, high-risk systems will face tough requirements regarding technical documentation and bias management. Joost van der Burgt, who leads AI supervision at the AP, suggested that organisations might be trying to avoid scrutiny.
Aleid Wolfsen, the chairman of the Dutch Data Protection Authority, referenced the childcare allowance scandal (toeslagenaffaire), in which algorithms wrongly accused citizens of fraud. He noted that although the lessons from five years ago are clear, action has been slow because solid rules and enforcement for AI are still missing.
@anp | NEWS BRAINPORT
