High-risk § 3.4.1

The personal scope of the use cases listed in point 4 of Annex III covers ‘employment, workers’ management and access to self-employment’³³. The AI Act does not itself define those notions, but requires a common understanding of them³⁴.

CJEU case law has developed an autonomous notion of ‘worker’ in relation to Article 45 TFEU, from which inspiration may be drawn to define the notion of employment. According to that case law, the essential feature of being a worker is that, for a certain period of time, a person performs services for and under the direction of another in return for remuneration³⁵. However, the notion of ‘worker’ must be considered as just one element of the broader expression ‘employment, workers’ management and access to self-employment’ and the reference in Recital 57 AI Act to ‘work-related contractual relationships’ should not be read limited to a formal employment contract s but as capturing the broader spectrum of arrangements that constitute work , ‘employment’, and self-employed activity in general (e.g., natural persons who are self-employed due to regulatory requirements or choice, but are formally integrated within an organisation to whom they provide services).

The notion ‘worker’s management’ recognises AI-enabled management of workers as a feature increasingly present in the workplace. Including workers’ management within the scope of the high-risk use cases listed in Annex IIIallows the AI Act to address the risk of harm to fundamental rights at the hiring and employment stage. The provisions of the AI Act for high-risk AI systems , in conjunction with the prohibition laid down in Article 5(1)(f) on emotion recognition in the workplace, establish a comprehensive legal framework for the regulation of the development of AI systems automating managerial tasks and a minimum standard for the protection of the rights of the workers in respect of the use of AI systems by employers³⁶. Managerial functions such as, inter alia, recruitment processes, work allocation, monitoringand worker evaluation, as well as decisions on promotion, remuneration or termination are covered by the use cases listed in point 4 of Annex III.

The inclusion of ‘access to self-employment’ in point 4 of Annex III aims to cover self-employed individuals such as persons working as independent contractors, service providers or performing independent professions rely on access to contracts or assignments to secure their livelihood. While Recital 57 AI Act recognises that work-related contractual also involve persons providing services through platforms³⁷ the notion of ‘ access to self-employment’ is broader than platform work and covers other forms of self-employed activity such as solo self-employed persons who are in a situation comparable to workers³⁸. In this regard, the EU acquis extends some protections to the self-employed, as regards the access to opportunities for professional activity³⁹. As a consequence, freelancers, independent professionals , service providers , and platform workers regardless of contractual status fall within the personal scope of the use cases listed in point 4 of Annex III wherever AI systems mediate or condition their access to work opportunities. The Commission Guidelines on prohibited artificial intelligence practices⁴⁰ further clarify that the notion of ‘workplace’ must be interpreted broadly, encompassing ‘any specific physical or virtual space where natural persons engage in tasks and responsibilities assigned by their employer or by the organisation they are affiliated to, for example in case of self-employment.’

Where a n AI system intended to be used in the field of employment falls within one of the prohibitions laid down in Article 5 AI Act, its placing on the market, putting into service and useis per se unlawful, irrespective of any safeguards put in place by its provider or deployer. By contrast, AI systems falling within one of the use cases listed in point 4 ofAnnex III are classified as ‘‘high-risk’ and subject to heightened compliance obligations. If an AI system falls within the scope of both provisions, the prohibition will prevail and the system will remain unlawful. In this context, it is particularly relevant to consider the prohibition concerning AI systems inferring emotions in the workplace, which are prohibited pursuant to Article 5(1)(f) AI Act. For examples of AI systems captured by the prohibition in Article 5 (1)(f) AI Act , reference is made to the Guidelines on prohibited artificial intelligence practices⁴¹.