The Council of the European Union has given final approval to the EU's landmark AI bill, setting the stage for the first standard Europe-wide AI law.
The EC explained that the AI Law takes a “risk-based approach” to AI regulation, meaning that the greater the risk that an AI product poses to harm society, the more regulations it will have to comply with to operate in the EU. At the same time, the EC said it hopes the AI Law will open the door to investment in the right kind of AI: safe and transparent AI.
As such, “high risk” AI systems – those used in sensitive sectors or by law enforcement agencies for purposes such as identification – will have to undergo regular audits, pass fundamental rights impact assessments and be registered in a central database of high risk AI systems.
“With the AI law, Europe will underline the importance of trust, transparency and accountability when dealing with new technologies, while ensuring that this rapidly changing technology can thrive and foster innovation in Europe,” said Belgian Digitalization Minister Mathieu Michel.
The AI Act classifies and regulates different types of AI and the companies that create them, and also declares some types of AI to be banned within the EU.
“Cognitive behavioral manipulation and social scoring will be prohibited in the EU as the risks are deemed unacceptable,” the European Commission said. “The law also prohibits the use of AI for predictive policing based on profiling, as well as systems that use biometric data to classify people according to certain categories, such as race, religion or sexual orientation.”
Potential future AGIs are also included in the regulations, and the act provides exceptions for systems used for military/defense and research purposes.
The law also calls for the creation of several new offices to implement the new law. AI offices at the commission and parliamentary levels will be established, as well as expert committees and stakeholder advisory forums. Fines for violations are also considered, which would be set as a “percentage” of annual global turnover or a pre-determined amount, whichever is greater.
The AI bill passed by the European Parliament in December has been criticized for appearing to be written with big tech companies in mind, but it's now passed and is set to have a significant impact when it comes into force in two years' time.
“Non-EU companies that use EU customer data in their AI platforms will have to comply with this law,” lawyer Patrick van Ecke told Reuters of the law's passage. “Other jurisdictions will likely use the AI law as a blueprint, just as they did with GDPR.”®
