The European Union is establishing an expert group comprising officials from member states to address queries on product safety, trade secrets, copyright, biometric identification, and law enforcement related to the recently approved AI rulebook.

The AI Act, which categorises AI systems into four categories based on potential risks to society, has been agreed upon by EU countries. However, concerns were raised by countries like France, Germany, and Italy regarding the regulation of specific AI systems, such as foundation models like ChatGPT. Germany supported the expert group, seeking clarity on the relationship between the AI Act and other legislation, particularly regarding law enforcement and border control systems. France wanted more clarity on trade secrets, while Hungary requested additional details on law enforcement.

The technical details of the AI Act have been agreed upon, but some provisions may require delegated acts or secondary legislation for further clarification. The approval of EU lawmakers is necessary before the act can be implemented, with a plenary vote in the European Parliament expected in April. Once approved, the act is set to enter into force later this year, with a 36-month implementation period. Requirements for AI models will start to apply within one year and prohibited systems will need to be phased out within six months. In addition to the expert group, the European Commission plans to establish four supervision and enforcement bodies, including the AI Office, European Artificial Intelligence Board, Advisory Forum, and Scientific Panel, to oversee and regulate AI technology.

Why does it matter?

This legislative framework, the first of its kind globally, categorises AI systems based on their potential societal impact, striking a balance between fostering innovation and addressing concerns such as product safety, trade secrets, and law enforcement. The harmonisation of regulations across EU member states and the creation of oversight bodies underscore the EU's commitment to shaping responsible and unified AI practices with far-reaching implications for global AI governance.