Overview of the European Union’s Artificial Intelligence Act.

1. European Union’s Artificial Intelligence Act

On August 1, 2024, the European Union’s (“EU”) Artificial Intelligence Act (“AI Act”) will take effect. The AI Act, the first comprehensive AI law confronts the development of AI, its marketing and use. As misinformation and disinformation spread throughout the globe, the EU states part of its fundamental values of creating the act is to protect democracies. 

1. Risk Categories

Risk, is divided into two categories, “unacceptable” and “high.” Unacceptable risk AI systems are considered a threat to people and will be banned by the EU. Notably, the EU regards social scoring as a prime example of unacceptable risk, clearly stating in the EU AI Act, and on the European Parliament’s website that  “unacceptable risk is prohibited (i.e.: social scoring).” A sharp contract to China’s view on social scoring. Other unacceptable risks include cognitive behavioral manipulation of people or specific vulnerable groups, such as voice-activated toys that encourage dangerous behavior in children (imagine if the famous horror doll Chuckie came to life…yikes), biometric identification and categorization of people, real-time and remote biometric identification systems, such as a facial recognition. One example of exceptions to these subcategories includes biometrics for law enforcement purposes.

High risk AI systems that negatively affect safety or fundamental rights are divided into their own two categories. The first category involves AI systems within products that fall under EU safety regulations. The second category requires AI systems to be registered in an EU database. This includes essential societal functions, such as management and operation of critical infrastructure, education, law enforcement, border control and assistance in legal interpretation and application of law. Each high-risk AI system will be assessed pre-market and throughout its lifecycle. A designated national authority will be available for the public to file complaints against high-risk systems. 

1. Agency Creation and Enforcement Structure

The European Commission created the European AI Office to implement the AI Act, providing a central base for all 27 EU Member States to collaborate with the European AI Office. The AI Act establishes the European Artificial Intelligence Board, comprising of the European Commission and representatives chosen by the Member States. On the national level, Member States choose authorities to work with their national supervisory authority, to oversee the implementation of the AI act.

1. Effective Dates

As mentioned above, the AI Act will enter into force on August 1, 2024. However, obligations pertaining to general-purpose AI governance will take place on August 2, 2025. A year later, August 2, 2026, all rules of the AI Act take effect. Finally, obligations for all other high-risk systems become applicable on August 2, 2027.

FAQs

1. What is the EU AI Act?
   1. The EU AI Act is the world’s first comprehensive law regulating the use of AI.
2. Are there different risk categories in the EU AI Act?
   1. Yes, notably unacceptable and high-risk. Each of which includes subcategories.
3. When does the EU AI Act take effect?
   1. The EU AI Act will take effect in stages, starting on August 1, 2024 and fully taking effect by August 2, 2027.