In this episode, we focus on the accuracy requirements under the EU AI Act for high-risk AI systems. Our discussion considers what "accuracy" means in the context of AI models, and different ways of thinking about how a model can be accurate or inaccurate in its outputs, as well as looking at different approaches that the Commission might take, in upcoming guidance, to specifying how the accuracy of AI models is to be measured. We also look at how the requirement for high-risk AI models to be sufficiently accurate is likely to play out in practice – examining the different obligations falling on providers and deployers of high-risk AI systems and what organisations are likely to need to do to meet these obligations.

In this episode, we turn our attention to the transparency obligations imposed on providers and deployers of limited and minimal risk AI systems under the EU AI Act. While these systems may not pose the same level of concern as high-risk AI, they are still subject to important requirements (particularly around informing users when they are interacting with AI, ensuring compliance in AI generated decision making around health, safety or basic rights). Our episode considers the practical implications for businesses developing or deploying AI in the EU.

In this episode, we examine high-risk AI systems under the EU AI Act and the specific exceptions that apply. We'll discuss the criteria defining high-risk categories, the compliance obligations for these systems, and the safeguards businesses must adopt. We also explore scenarios where certain high-risk requirements may not apply, clarifying how exceptions function within the Act, and detailing the different obligations for different roles in the AI supply chain. Join us to understand how businesses can approach AI risk thoughtfully and remain compliant with evolving regulations.