Dutch Citizens ‘Should Not Expect Miracles’ From AI Act, Says Dutch Privacy Agency
The Dutch privacy regulator said imminent artificial intelligence regulation in the European Union may fail to prevent the rollout of dangerous algorithms.
Europe is close to finalizing the AI Act and enacting regulations intended to mitigate societal risks and ban a slew of applications, such as biometric recognition in public places, that are deemed harmful to society (see: Europe Closes in on Rules for Artificial Intelligence).
Dutch data protection authority Autoriteit Persoonsgegevens said the pending legislation may fall short and told citizens of the Netherlands that “we should not expect miracles from the AI Regulation.”
The data protection authority said the regulation depends on industry self-assessments for high-risk applications. “As a result, there is an explicit possibility that unsuitable high-risk algorithmic systems will enter the market and be used by private and public parties.” By the time that a regulator succeeds in removing the application from the market, “the damage may have been done.”
The Dutch government itself has courted controversy over its unregulated use of AI algorithms, and Amnesty International in 2021 accused the government of using an algorithm that racially profiled claimants for childcare benefits with false allegations of fraud.
The warning from the Dutch agency comes as part of its AI risk assessment report published last week by the new Directorate of Algorithm Coordination within Autoriteit Persoonsgegevens.
The reports calls on the national government to direct developers to complete an AI risk assessment on new systems prior to implementation. It also recommends that by early next year the government should set up its own algorithm public repository to identify high-risk AI systems used in the country.
The directorate “is also in the process of developing a perspective on algorithm risk,” the agency wrote.
In anticipation of the EU artificial intelligence regulation, other trade bloc members have also announced their plans to supervise AI at a national level. Last month, the Spanish government announced Europe’s first-ever dedicated artificial intelligence regulatory agency. Earlier, the French data protection agency said it will release its four-pronged action plan to promote privacy-friendly AI systems.