Artificial Intelligence (AI) has been the subject of intense debate and scrutiny over the past few years. The European Union (EU) has recognized the need for legislation to regulate AI to ensure that it does not cause harm to individuals or society as a whole. The EU's AI Act is a legislative proposal that seeks to regulate AI based on its capacity to cause harm. One of the most politically sensitive parts of the file is the definition of AI, which will define the scope of the EU's AI rulebook.
 |
| EU Adopts OECD Definition for AI in AI Act |
The definition overlaps largely with the one used by the OECD, an international organization that is often regarded as a club for rich countries. The EU Parliament added a preamble to the text, calling for the AI definition to be closely aligned with the work of international organizations working on artificial intelligence to ensure legal certainty, harmonization, and wide acceptance.
Conservative lawmakers from the European People's Party wished to narrow the definition to systems based on machine learning, while left-to-center lawmakers wanted a broader approach targeting automated decision-making. However, the compromise notes that where an AI model is integrated into a more extensive system entirely dependent on the AI component, the entire system is considered part of a single AI solution.
The compromise also specifies that AI should be distinguished from simpler software systems or programming approaches, and the set objectives could be different from the intended purpose of the AI system in a specific context. The wording has been included stating that the reference to predictions includes content, a measure intended to ensure that generative AI models like ChatGPT do not fall through the cracks of the regulation. However, the next question that arises is whether such models fall under high risk and chapter 2 or can count on a big GPAI exemption.
The EU's adoption of the OECD's definition of AI is a step towards harmonization and legal certainty, but it remains to be seen how the EU will regulate AI to prevent harm.
EU Agrees on Other Definitions for the AI Regulation
On March 6, 2023, an agreement was reached at a technical meeting on most other definitions of the AI regulation. The EU Parliament agreed to define "significant risk" as a risk that is significant in terms of its severity, intensity, probability of occurrence, duration of its effects, and its ability to affect an individual, a plurality of persons or to affect a particular group of persons.
The EU Parliament also defined remote biometric verification systems as AI systems used to verify the identity of persons by comparing their biometric data against a reference database with their prior consent. That is distinguished by an authentication system, where the persons themselves ask to be authenticated. The definitions of biometric categorization were also clarified, with a reference added to inferring personal characteristics and attributes like gender or health.
The EU Parliament introduced a reference to the Law Enforcement Directive for profiling by law enforcement authorities. The EU is working to ensure that AI does not harm individuals or society as a whole. The EU's AI Act is a legislative proposal that seeks to regulate AI based on its capacity to cause harm. These definitions will play a critical role in defining the scope of the EU's AI rulebook and how AI is regulated in the EU.