The practical implementation of the AI regulation continues to take shape. The EU Commission has presented the long-awaited draft guidelines for the classification of AI high-risk systems, including comprehensive annexes, as part of a public consultation. The initiative aims to shed more light on the AI Act and ensure consistent application and effective enforcement of Article 6. This forms the basis for risk classification.
Read more after the ad
The guidelines are intended as a guide for the national market surveillance authorities, which in Germany include the Federal Network Agency. At the same time, they provide developers, providers and operators of AI technologies with a tool to legally determine whether their applications fall into the sensitive high-risk category.
The Commission emphasizes that it is a fundamental yes-no question as to whether a system should be classified as high-risk. She does not see the practical examples listed as an exhaustive list, but rather as a dynamic document.
Flexibility and focus on fundamental rights
The EU Commission reserves the right to continually update the examples in order to adapt them to technological developments. Further guidelines that regulate in detail the specific compliance with the requirements as well as specific obligations for providers and operators, for example when labeling chatbots and deepfakes, are already being prepared. The Commission explains that it deliberately limited the scope of the high-risk term and made it proportional. It focuses strictly on systems that pose a significant risk to health and safety or have a noticeable negative impact on the fundamental rights of citizens.
The systematic classification is based on two pillars. Article 6(1) covers AI systems that are either themselves subject to the European harmonization rules set out in Annex One as a product or are used as a security component in such a product. The prerequisite for this is that the end product must undergo a mandatory conformity assessment by a third party.
The guidelines provide important regulatory elements and an evaluation methodology for this area. The focus is not on the individual product, but rather on its occurrence in the existing list of harmonization regulations, which can also be modified in the future.
Smartwatches in the sights of regulators
Read more after the ad
The second pillar concerns Article 6(2). This refers to standalone systems identified as high risk in Schedule Three of the AI Act. To illustrate this, the Commission divides this area into eight overarching categories, including critical infrastructure, education and biometrics.
An example from the area of emotion recognition is likely to provide material for discussion in the wearable industry. The draft stipulates that AI systems in smartwatches that are used to identify feelings and track biometric data such as heartbeat should be classified as high-risk applications.
To make it easier for European industry to adapt to the new standards, the EU is giving companies more time. As part of the AI omnibus, the original deadlines were pushed back: the strict obligations for AI systems under Article 6 paragraph 2 apply from December 2, 2027, while for products under paragraph 1 the grace period even applies until August 2, 2028.
Binding core requirements
With regard to harmonization, the Commission clarifies that “the classification of an AI system as a high-risk system in accordance with Article 6(2) will have a uniform effect across all Member States”. The point is to adhere to clear minimum standards, not a ban. The Brussels lawyers emphasize that these systems are “subject to appropriate requirements to ensure that they function accurately and as intended and that risks to health, safety and fundamental rights are properly assessed and mitigated”.
The draft also closes loopholes for developers. A provider cannot therefore “exempt an AI system from regulation and classify it as ‘low risk’” by “adding a requirement for human intervention”. The exceptions to the filter mechanism anchored in the law are not a free pass either: “The conditions must be interpreted narrowly.”
(wpl)
