Italian Regulations on Artificial Intelligence Hiring

Employers are more and more utilizing Artificial Intelligence (AI) within the hiring course of. For instance, chatbots can reply candidates’ questions, instruments can display screen resumes and profile candidates, and software program can rating interviews.

As employers proceed to discover how AI can streamline processes and reduce bias, they need to pay attention to the bounds of AI, their authorized obligations when utilizing AI, and guiding ideas relating to the implementation of AI.

In Italy, these authorized obligations and guiding ideas embrace Legislative Decree no. 104/2022¹ (the Transparency Decree), the Code of Conduct for Staff Leasing Agencies² (the Code of Conduct), which was authorised by the Data Protection Authority’s measure no. 12/2024,³ Italy’s Artificial Intelligence Bill⁴ (the AI Invoice), which is at the moment below overview by the Italian Parliament, and the European Union Artificial Intelligence Act⁵ (the EU AI Act).

This alert will focus on employer obligations below the Transparency Decree, Code of Conduct, and AI Invoice, and can present suggestions to employers as they navigate the ever-changing panorama of AI use in hiring. For extra data on the EU AI Act, please try Okay&L Gates’ supplies on the EU AI Act, out there here,⁶ here,⁷ and here.⁸

I. The Transparency Decree

In August 2022, the Transparency Decree took impact and, consequently, Directive (EU) 2019/1152,⁹ also called the Directive on Clear and Predictable Working Circumstances, was applied in Italy. The Transparency Decree addresses automated decision-making instruments and monitoring techniques. Relating to the usage of AI in employment, the Transparency Decree requires employers to supply their workers, job candidates, any commerce union representatives inside the firm, and, if no such commerce union representatives exist, the territorial commerce unions’ our bodies with details about:

What features of the employment relationship could also be affected by AI.
The aim and operation of the AI instruments in use.
What information and parameters are used to coach the AI.
What management measures, corrective measures, high quality management techniques, and cybersecurity instruments are in use.

After the Transparency Decree took impact, the Italian Ministry of Labour and Social Insurance policies (Ministry of Labour) issued a round to make clear the decree’s scope. The Italian Ministry of Labour’s 20 September 2022 circular (Round No. 19)¹⁰ defined that obligations below the Transparency Decree apply if there’s solely incidental human intervention when implementing the AI instrument and supplied examples of instances in which there’s an obligation to supply data below the Transparency Decree, comparable to:

Recruitment or project utilizing chatbots through the interview, automated profiling of candidates, screening of resumes, and use of emotional recognition software program and psycho-aptitude assessments.
Administration or termination of employment with automated project or revocation of duties, duties or shifts, definition of working hours, productiveness evaluation, willpower of wage, promotions, and so forth. via statistical evaluation, information analytics or machine studying instruments, neural networks, and deep studying.

Round No. 19 additionally reminded employers that they need to make sure that their use of AI (in addition to different decision-making instruments or monitoring techniques) complies with Article 22 of the General Data Protection Regulation (GDPR),¹¹ together with by performing a threat evaluation and an affect evaluation of the processing actions carried out.

In March 2023, the Court docket of Palermo issued the primary judgment in relation to the Transparency Decree (Court of Palermo, judgment no. 14491/2023),¹² discovering that the employer violated the Transparency Decree when it failed to speak to the related commerce unions the standards for use as a part of the functioning of the AI algorithm. The ruling serves as an vital reminder to employers to familiarize themselves with their obligations with regards to utilizing AI in employment.

II. The Code of Conduct

In January 2024, Italy’s Knowledge Safety Authority issued the Code of Conduct.¹³ Beneath the GDPR, codes of conduct are voluntary units of guidelines that help the issuing member with information safety compliance and accountability in particular sectors or referring to specific processing operations. Codes of conduct should be authorised and monitored by the member state’s GDPR supervisory authority, right here, Italy’s Knowledge Safety Authority.

This Code of Conduct gives a set of greatest practices for employees leasing companies dealing with staff’ and candidates’ private information, together with by addressing the usage of AI in choice and recruitment processes. Particularly, the Code of Conduct explains that:

Employees leasing companies can use automated techniques within the choice and recruiting course of as long as they perform an in depth affect evaluation and supply staff and candidates with clear data as to AI’s periodical critiques and mechanisms.
If absolutely automated techniques are used, staff ought to be permitted to a minimum of get hold of human intervention, categorical their opinion concerning the choice, and problem the choice.

Though the Code of Conduct solely applies to employees leasing companies and isn’t binding, it incorporates useful data for all employers that can be utilized to adjust to present authorized obligations that implicate AI, together with GDPR.

III. The AI Invoice

Lastly, to make sure they don’t run afoul of AI-related guidelines sooner or later, employers ought to familiarize themselves with Italy’s AI Invoice and take steps to organize for its passage. Within the employment context, the AI Invoice:

Reaffirms the precept of equity and nondiscrimination in the usage of AI.
Asserts that AI should be used to enhance working situations, defend the psychological and bodily integrity of workers, and enhance the standard of labor efficiency and productiveness of individuals in accordance with EU regulation.

The AI Invoice additionally specifies that the usage of AI within the office should be protected, dependable, and clear, and can’t have an effect on human dignity or breach confidentiality of non-public information.

IV. Suggestions

In mild of the prevailing and anticipated legal guidelines, steerage, and directives relating to the usage of AI in sourcing and hiring expertise, employers ought to take into account taking the next steps when assessing the advantages and implementation of AI instruments in hiring:

Perceive what the AI is doing for you and the danger ranges related to these features. Is it a high-risk instrument as a result of it’s making the choice for you and, thus, lacks human intervention at key factors within the course of (e.g., resume screening instruments, gamified assessments, AI-scored interviews)? Is it a medium-risk instrument as a result of it’s suggesting a call and, thus, has some human involvement and oversight at key factors (e.g., rating candidates, pushing on-line job postings to sure folks)? Or is it a lower-risk instrument as a result of it’s creating content material and, thus, permits for human overview and oversight earlier than supplies are finalized (e.g., drafting job postings, insurance policies, or textual content summaries of conferences)?
Commonly audit the instrument to determine potential algorithmic bias and to guage the instrument’s outcomes and implement corrective measures as vital, together with by participating outdoors consultants to do a validation research.
Create a process power or inner governing physique to coordinate inner oversight and moral tips.
Implement an organization coverage on the usage of AI, conduct periodic coaching on AI use, and solely allow these with correct coaching to make use of the AI instrument and approve any AI-suggested employment choices.
Present workers and job candidates with a correct written data discover stating the AI’s guidelines, functions, and mechanisms.
Present commerce union representatives (or related territorial commerce unions’ our bodies) with a correct written discover stating the AI’s guidelines, functions, and mechanisms.
Embody human overview within the job software course of (e.g., in-person interview with out electronics).
Embody human validation of the ultimate choices and outcomes given by AI.
Be certain that personal and delicate job applicant and worker data utilized by AI is shielded from improper disclosure and that the AI instrument in any other case complies with privateness and information safety rules, together with GDPR.
Ought to the AI additionally allow monitoring of the actions of your workers, make sure that an settlement along with your commerce unions’ representatives is reached upfront, or—of their absence—that the formal authorization by the related district labor workplace is correctly given.
Be certain that any AI distributors you utilize have thought of and brought steps to fight algorithmic bias, together with via criterion validity research; make sure that the seller contracts can help you entry and, if vital, disclose the seller’s information; and focus on with distributors what would occur if there was a authorized violation, discovering of algorithmic bias, or different situation with their instrument.