The European Union’s (EU) new legislation on artificial intelligence (AI) is experiencing revised compliance deadlines due to a delay in its publication, despite the text being finalized.
EU officials have informed Osborne Clarke that the publication of the definitive text in the EU’s Official Journal, originally anticipated earlier, is now postponed until mid-July, potentially around 12 July 2024. Consequently, the AI Act is expected to become legally binding approximately 20 days after this publication, likely in early August.
This delay grants businesses additional time to grasp forthcoming compliance responsibilities and guidance from the AI Office. As a result of the publication delay, the implementation and compliance timelines have been adjusted accordingly:
- Early 2025: Prohibitions on unacceptable risks and general provisions on subject matter, scope, definitions, and AI literacy take effect.
- Late summer 2025: Provisions regarding general-purpose AI, governance, conformity bodies, enforcement, confidentiality, and penalties come into force.
- Late summer 2026: All remaining provisions, including regulation of high-risk AI systems, are implemented.
- Late summer 2027: Regulation of high-risk systems within the EU product safety regime specified in Annex I takes effect.
The AI Office, established as a unit within the European Commission, has already commenced operations as the EU’s center of AI expertise. It will play a pivotal role in the AI Act’s implementation, providing guidelines and codes of practice as mandated by the legislation. Additionally, it serves as a platform for businesses to discuss interpretations and applications of the AI Act’s provisions concerning specific AI systems.
In light of these developments, businesses should also consider various other legal and regulatory risks associated with the development and deployment of AI systems, detailed in our comprehensive overview on AI-related considerations for businesses.
Source: osborneclarke.com
