Italy: Employer Investigations in the Era of Artificial Intelligence

published on 25 February 2025 | reading time approx. 9 minutes

The use of artificial intelligence in hiring processes raises new ethical and legal challenges, requiring an evolving interpretation of the prohibition of opinion investigations and particular attention to workers' fundamental rights.

Brief Overview of the Scope of Article 8 of the Workers' Statute

Article 8 of the Workers' Statute, titled "Prohibition of Opinion Investigations", prevents employers from conducting, either directly or through third parties, inquiries into the political, religious, or trade union opinions of workers, as well as any other aspect not relevant "for assessing professional aptitude". The provision establishes a dual prohibition applicable both during the pre-employment phase and throughout the employment relationship: the first, absolute and instrumental to the provisions of Article 15 of the Workers' Statute on discriminatory acts, concerns investigations into workers' political, religious, or trade union opinions; the second has a broader scope, extending to any fact irrelevant to assessing the worker's professional aptitude.

While the first prohibition - explicitly concerning only political, religious, or trade union opinions - does not raise interpretative doubts, the same cannot be said for the second prohibition, whose scope is not entirely clear, making it difficult to determine with certainty when a fact can be considered relevant for assessing professional aptitude.

The previously held view that the concept of "professional aptitude" was limited to the worker's ability to perform specific tasks has been superseded. It is now widely accepted that this assessment may extend to several conducts affecting the employment relationship. However, significant challenges may be faced in precisely defining the scope of the prohibition. The general nature of the provision requires careful case-by-case interpretation, considering the specifics of each situation and the potential effects of a conduct on the proper performance of the employment relationship.

As anticipated, case law has progressively adopted a more expansive approach to employer investigations. Notably, rulings on recruitment processes and hiring procedures have granted the possibility of evaluating extraneous facts, provided they compromise the trust relationship underlying the employment contract or damage the company's image and economic interests. For instance, Cass. 14 August 2020, No. 17167 deemed legitimate a hiring process requiring candidates to present a certificate of pending charges.

Article 8 of the Workers' Statute and the Test of New Technologies: The Impact of Artificial Intelligence

With the advent of artificial intelligence systems the risk is even greater, as these tools use advanced algorithms to process large amounts of data, allowing employers to collect, analyze, and cross-reference personal information - often without the explicit consent of employees - thus expanding the possibility of investigating aspects of their private lives, including opinions and behaviors, in a potentially invasive and privacy-infringing manner.

This risk is exacerbated by the difficulty in monitoring and limiting the misuse of these technologies, whether due to a lack of familiarity or disregard for their potential. Such technologies could generate indirect discrimination or non-transparent evaluations, conflicting with the prohibitions and guarantees provided by Article 8 of the Workers' Statute.

Although some legal scholars argue that the concept of "investigation" should be redefined and updated in light of new technologies, incorporating information collected and processed by artificial intelligence, the current wording of Article 8 of the Workers' Statute is already broad enough to encompass any form of knowledge-gathering activity through which an employer, even with the aid of artificial intelligence, collects information and thus conducts an investigation on prohibited matters.

The Use of Artificial Intelligence Systems in Hiring: Focus on Social Recruiting

A particularly sensitive issue within the scope of Article 8 of the Workers' Statute is the use of algorithms and artificial intelligence in the hiring process. While these technologies simplify and expedite decision-making, they also introduce various ethical, legal, and managerial concerns.

Specifically, the risks related to the invasiveness of candidate data processing, the purposes of such operations, and the reliability of algorithmic decisions raise significant issues, which will only increase with the growing use of these tools and the limited presence of case law precedents on the matter.

One of the most revolutionary phenomena in recruitment has been the introduction of social media and social networks into the job market (so-called "social recruiting"). Platforms such as LinkedIn and Indeed have become essential for companies seeking talent (and for job seekers). Additionally, general-use social media, like Facebook, are increasingly used for professional purposes, further expanding recruitment possibilities and data collection.

Employers use these tools for various purposes, including job postings, candidate searches through online professional profiles (and sometimes non-professional ones), verification of information provided by candidates in interviews or resumes, and analysis of their web reputation. However, these practices raise serious concerns regarding the reliability of information obtained from social media. The ease with which individuals can present themselves online in ways that do not fully reflect reality can compromise the quality of hiring decisions. This results in a high risk of bias in selection processes, potentially leading to unfair or incorrect assessments of candidates.

Regarding social media usage, candidate profiling through algorithmic matching models enables automated evaluations of their potential and compatibility with certain work environments, as well as predictions about their future behavior.

Through artificial intelligence systems, data on candidates' private lives - such as daily habits, personal relationships, or social media preferences - can be collected. An emblematic example could involve a candidate posting a vacation photo with ironic comments about their relaxed lifestyle. Similarly, images depicting political or trade union events, or even posts commenting on current events or politics in colorful language, could be analyzed.

All these behavioral data points can be used to build a personality profile or even predict future workplace behavior. Another area of concern is biometric data, such as facial recognition or expression analysis during video-recorded interviews. These tools, already used in the United States, could pose a disturbing invasion of personal privacy, especially when none of the legal bases for processing special categories of data (particularly consent) under Article 9 (2) of Regulation (EU) 2016/679 (GDPR) are met. This is particularly concerning when adequate information is not provided, impact assessments are not conducted, or technical measures such as encryption and immediate data deletion are not implemented.

Although these tools offer undeniable analytical potential, their reliability is not absolute. Human oversight is essential in designing system models and defining parameters (see Article 14 of the AI Act). The risk remains that algorithms could be influenced by the opinions of their programmers, statistical distortions, or social stereotypes, thereby generating the aforementioned biases.

Consequently, while artificial intelligence is destined to become a fundamental resource in managing recruitment and selection processes, its use requires careful and responsible handling.

Notably, the recent Regulation (EU) 1689/2024 (AI Act) classifies AI systems used "for hiring or selecting individuals" as high-risk under Article 6(2) if they pose a significant risk to health, safety, or fundamental rights. These AI systems (so-called stand-alone systems), considered high-risk, are listed in Annex III of the Regulation, which explicitly includes, among others, tools used to publish targeted job advertisements, analyze or filter applications, and assess candidates.

However, the AI Act itself provides for exceptions. For example, certain AI solutions would be exempt from the restrictions imposed on high-risk systems if used to improve the outcome of an already completed human activity or if employed for preliminary tasks leading to a more complex and significant assessment. Essentially, the exemption applies to systems that perform preparatory tasks or merely support the human decision-making process without directly influencing the final evaluation of the candidate.

For AI systems used in recruitment, it is crucial that data processing is strictly limited to what is essential for evaluating job performance. This requires adopting an approach that minimizes the use of personal data, ensuring its relevance to the purpose of the processing.

The screening and selection of candidates' résumés, especially when conducted through artificial intelligence systems, necessarily involve the processing of personal data, thus requiring compliance with the GDPR.

As is well known, the GDPR mandates that any company or organization collecting and analyzing personal data - such as in the case of recruitment - must adhere to strict principles of lawfulness, fairness, and transparency. Additionally, it must ensure the adoption of adequate measures to protect the fundamental rights and freedoms of candidates, preventing potential discrimination arising from profiling processes or automated decisions. These measures may include risk assessments, impact evaluations, information notices, consent forms, and technical and organizational safeguards.

In line with the principle of privacy by design, the AI system must operate using only relevant and up-to-date information to verify the worker's suitability for the specific job, while fully ensuring the confidentiality and rights of the individual.

Final Considerations

Article 8 of the Workers' Statute is a fundamental pillar for protecting dignity and individual freedom in employment relationships, safeguarding not only the worker's private sphere but also what Privacy Commissioner Stefano Rodotà referred to as the "public projection of the person". As Rodotà himself stated, "Data [...] are not confined to a strictly private sphere from which any unwanted gaze must be excluded. Instead, they constitute the public projection of the person, so that accompanying them with a prohibition on collection serves to prevent a worker from being discriminated against on this basis".

In a context increasingly shaped by the use of artificial intelligence systems and, more broadly, automated tools for recruitment, it is crucial to ensure that such technologies are employed in accordance with the principles of transparency, proportionality, and data minimization. Workers must not only be informed about the purposes and logic of the adopted systems but must also be able to exercise their right to access processed data and decision-making logic, either directly or through trade union representatives.

Despite the challenges posed by technological advancements, the Italian legal framework appears sufficiently equipped to address these changes, thanks to the adaptability of its regulations, foremost among them Article 8 of the Workers' Statute.