The Slovenian Presidency of the Council of the European Union ('the Council') published, on 29 November 2021, its compromise text on the Proposal for a Regulation of the European Parliament and of the Council Laying Down Harmonised Rules on Artificial Intelligence ('the AI Act'). In particular, the changes introduced by the Council pertain to, among others, provisions on:

• the subject matter and scope of the AI Act;
• various definitions;
• prohibited AI practices;
• classification rules for high-risk AI systems;
• the general purpose of AI systems; and
• delegation of powers to the European Commission and reporting.

Notably, the proposed changes include a fully redrafted defintion of an AI system. Specifically, the Council highlighted that the amended definition of an AI system is intended to ensure more legal clarity and to better reflect what should be understood by an AI system, with an explicit reference indicating that any such system should be capable of determining how to achieve a given set of human defined objectives by learning, reasoning, or modelling, and also aims to prevent the inclusion in the scope of the AI Act of more traditional software systems that are normally not considered as AI. 

In addition, the proposed changes provide that 'personal data' within the meaning of the AI Act means personal data as defined under the General Data Protection Regulation (Regulation (EU) 2016/679) ('GDPR'), whilst non-personal data means data other than personal data defined under the GDPR. Among other additional changes, the compromise text seeks to remove AI systems used for national security purposes from the scope of the AI Act and to extend the prohibition of using AI for social scoring also to private actors.

You can read the compromise text here.