Europe’s Battle to Control High-Risk AI: Hurdles and Resolutions

by | Sep 11, 2023

The European Union (EU) is currently facing a tough task in effectively regulating high-risk artificial intelligence (AI) systems as they become more common. Proposed laws aimed at limiting dangerous algorithms may not be enough to achieve their goals, raising concerns about the potential emergence of unsuitable and racially biased applications in the market. However, European countries are taking steps to address these challenges at a national level, with the Netherlands and France leading the way.

In the Netherlands, the unregulated use of AI algorithms by the government has faced intense scrutiny. Amnesty International has accused the authorities of using an algorithm that racially profiles claimants for childcare benefits. Such instances highlight the urgent need for strong regulation to prevent AI systems from perpetuating discrimination and bias. Unfortunately, doubts have been raised about the effectiveness of the pending laws, as they rely on industry self-assessments for high-risk applications.

The Dutch data protection authority has cautioned citizens not to expect miracles from the AI Regulation, emphasizing the importance of thorough assessment and oversight. Experts argue that while the laws aim to address the risks associated with high-risk algorithmic systems, there is a possibility that unsuitable applications may still enter the market and be used by both private and public entities.

In recognition of the need for a proactive approach, the Spanish government has taken a significant step forward by announcing the establishment of Europe’s first-ever dedicated artificial intelligence regulatory agency. This move underscores the recognition of the crucial role thorough oversight plays in ensuring the responsible use of AI technologies. Similarly, the French data protection agency is preparing to release a four-pronged action plan to promote privacy-friendly AI systems, highlighting the importance of directly addressing algorithmic risks.

Meanwhile, in the Netherlands, the Directorate of Algorithm Coordination within the data protection authority has published an AI risk assessment report. The report recommends the creation of an algorithm public repository, serving as a centralized platform for identifying and monitoring high-risk AI systems used within the country. This repository would play a vital role in identifying and addressing potential risks associated with AI applications.

However, concerns have been raised about the potential harm caused by high-risk AI systems before regulators can effectively remove them from the market. By the time such applications are identified and addressed, the damage may already be done. This emphasizes the urgency of implementing thorough risk assessment protocols before deploying AI systems.

Recognizing the limitations of the pending laws, other EU member states are also considering the option of supervising AI at a national level. This approach allows for more tailored regulation that takes into account the specific challenges faced by individual countries.

As Europe grapples with the complexities of regulating high-risk AI systems, policymakers must strike a delicate balance between fostering innovation and ensuring ethical and responsible usage. While the pending laws may serve as a starting point, they must be complemented by strong oversight mechanisms and proactive risk assessment practices. By addressing concerns surrounding algorithmic risks and promoting privacy-friendly AI systems, Europe can pave the way for a more responsible and accountable AI landscape.

In conclusion, Europe finds itself at a critical point in effectively regulating high-risk AI systems. The challenges are significant, but with the right combination of strong oversight, proactive risk assessment, and tailored national regulations, Europe can lead the way in promoting responsible and ethical AI usage. The time to act is now, ensuring that innovation continues to thrive while safeguarding against the potential risks associated with AI technologies.