The European legislation prohibits the use of artificial intelligence systems for evaluating individuals based on social behaviour or personal characteristics, where this involves processing that is prejudicial or discriminatory. Regulation (EU) 2024/1689 of 13 June 2024 (commonly referred to as the “AI Act”) recognised the potential ability of artificial intelligence systems to assign social scores to individuals (social scoring). It imposed restrictions to prevent this practice from producing discriminatory effects. Specifically, as specified in recital (31) of the AI Act, social scoring consists of the use of artificial intelligence technologies to evaluate and profile individuals based on a score obtained from a combination of data relating to: The EU legislator has noted that the above-mentioned practice risks constituting prejudicial or unfavourable treatment for the data subjects. Therefore, Article 5, paragraph 1, letter c) of the AI Act prohibits the use of social scoring practices when they lead to: Moreover, the European Commission has published Guidelines (C 2025/884) providing more specific indications on prohibited social scoring practices. Conversely, the legislator allows the use of such technology to evaluate individuals for a specific objective that is consistent with Union law (as specified in Recital 31 of the AI Act). Furthermore, paragraph 175 of the Guidelines clarifies that not all social scoring practices are prohibited, but only those which cumulatively meet the criteria set out in Article 5(1)(c) of the AI Act, rendering the practice capable of producing harmful or detrimental effects. In the context of human resources management, considering the Guidelines mentioned above, the use of social scoring appears to be permitted for conducting specific assessments of employees. The use of social scoring systems must in any case comply with the principles laid down in Article 5 of Regulation (EU) 2016/679 on the processing of personal data, as well as with the rules on automated individual decision-making and profiling (pursuant to Recitals 71 and 72 of the GDPR and the guidelines of the European Data Protection Board). Where the evaluation or classification is based on one of the characteristics protected under anti-discrimination law (such as age, religion, racial or ethnic origin, sex, etc.) or results in direct or indirect discrimination on such grounds, the practice will also fall within the scope of Union and national anti-discrimination legislation. Where artificial intelligence systems are intended to be used to assign a social score aimed at optimising employee evaluation—particularly where such scoring may determine the attribution of professional advantages or disadvantages—care shall be taken concerning the type of information contributing to the score, ensuring compliance with the aforementioned prohibitions, to avoid the risk of incurring the severe penalties provided for under the AI Act. In any case, the legitimate use of a social scoring system in the corporate context for employee evaluation constitutes the use of a high-risk AI system, as per Article 6 of the AI Act. Therefore, it will be necessary to conduct a risk assessment related to this practice and implement a risk management system in accordance with applicable European and national legal provisions.
2026 - Morri Rossetti
