The European Union (EU) has reached a compromise on a comprehensive regulation for artificial intelligence (AI), making it the first region in the world to comprehensively regulate the technology. The regulation still requires final approval from the European Parliament and Council, but this is typically considered a formality. The penalties for non-compliance can result in fines of up to 7% of global revenue. The EU’s regulations on AI are expected to have a global impact, as the region often sets the standard for cutting-edge regulations. In contrast, the United States has yet to approve a comprehensive AI regulation and relies on executive orders and voluntary commitments.
The EU regulation bans certain AI applications, such as social scoring, and takes a tiered-based approach to high-risk systems. It also prohibits mass scrapping of images from the internet for facial recognition algorithms. The compromise bill allows real-time biometric recognition in public with prior judicial authorization for crime-related searches. Parliament succeeded in imposing guardrails for foundation models, requiring developers to conduct evaluations, assess risks, conduct adversarial testing, report incidents, ensure cybersecurity, and report on energy efficiency.
However, the compromise has received criticism from European Digital Rights for its perceived loopholes and omissions. The group argues that AI developers can evade regulation if they report their systems falling below the EU’s high-risk threshold. It also raises concerns about the ban on emotional recognition systems, which it believes should include policing and border and migration contexts. Some European privacy regulations are concerned that dangerous algorithms may gain traction in society before they can be stopped.
Enforcement of the AI Act is expected to take time, similar to the General Data Protection Regulation (GDPR). Some European countries have already begun setting up national AI authorities to enforce the regulation. Critics of GDPR enforcement argue that challenges and court rulings are still ongoing for matters such as commercial data transfers to the United States and targeted advertising by online platforms.