The EU AI-Act went live during the European summer break. That's inline with the future plans of use of AI anyway: Have AI do the work and sit on the beach…

Here is the summary of the most relevant dates that organizations should have in their calendar (see full details in Art. 113 AI-Act): 

• August 2024: 
  • EU AI-Act in effect 
• February 2025: 
  • Prohibition of unacceptable AI practices 
• August 2025: 
  • Provisions for general purpose AI models contained in Chapter V AI-Act
  • Sanctions regulated in Chapter XII AI-Act
  • Appointment of Member State competent authorities
• August 2026: 
  • Full application of most rules of the AI Act - for example, rules on high-risk AI systems specifically listed in Annex III (e.g. biometrics, critical infrastructure, personnel management, law enforcement, migration, asylum and border control)
  • Member state authorities established  AI regulatory sandbox
  • Commission review, and possible amendment of list of high risk AIs
• August 2027: 
  • Obligations on product related high risk AI comprised by Annex I 

What organizations now need to do next: Set up their AI governance framework. This should start with allocating responsibilities amongst departments and employees within the organization and drawing up a “plan to governance”. We will follow up with details in our next post.