Bufete Barrilero (Spain) Article: AI Bill – Updates on Labor Law

On April 23, 2024, the Italian government presented a bill titled “Regulations for the Development and Adoption of Artificial Intelligence Technologies.”

While European AI regulations primarily focus on governing the design and market introduction of AI systems—considered as products—the Italian bill outlines general principles, goals, and objectives regarding AI usage, regulating its application across various sectors, including labor law.

Before delving into the specifics of the bill, it’s important to note that labor law is one of the first areas where judicial precedents in AI are already emerging. Some early AI-related case law includes:

  • – Bologna, Order No. 2949/2022: One of the first rulings concerning workplace discrimination due to AI systems.
  • – Palermo, Judgment No. 14491/2023: This ruling found that the employer’s failure to communicate with unions about the functioning criteria of an AI system used to assign rider jobs was anti-union behavior.
  • – Turin, Judgment No. 743/2023: Here, the criteria used by an AI system employed by the Ministry of Education to select substitute teachers were deemed unlawful.

AI and Labor Law in the Italian Bill

The AI Bill establishes the need for a human-centric approach to AI, emphasizing that AI should serve humanity.

In labor law, this concept translates into AI being used for the benefit of workers, aimed at improving working conditions, enhancing their potential, and protecting their rights.

The bill introduces its own definition of “AI models” (distinct from the AI Act’s definition of “general-purpose AI models” in Article 3, paragraph 1, No. 63). It defines AI models as those that “identify recurring patterns using data sets, capable of performing a wide range of distinct tasks, and that can be integrated into various systems or applications.” A practical example of such a model is “Chat GPT.”

It’s worth noting that the difference between the AI Act’s “general-purpose AI models” definition and the Italian bill’s “AI models” definition could potentially cause harmonization issues between European and national regulations.

Article 10: Provisions on AI Use in Labor

This provision focuses specifically on labor, regulating AI usage in this sector with the aforementioned human-centric approach. AI must:

  1. Improve working conditions,
  2. Protect the physical and mental well-being of workers, and
  3. Enhance the quality of work performance and productivity.

Additionally, AI’s use in the workplace must be safe, reliable, and transparent, while ensuring workers’ dignity and privacy.

Recognizing the importance of transparency, the Italian legislator mandates comprehensive information obligations for employers, expanding the requirements outlined in Article 1 bis of Legislative Decree 152/1997 (“Additional Information Obligations in the Case of Automated Decision-Making or Monitoring Systems”)—a regulation already established in the legal system.

Finally, the bill guarantees full respect for workers’ inviolable rights, aiming to prevent discrimination based on gender, age, ethnicity, religion, sexual orientation, political opinions, and personal, social, or economic conditions.

This provision is aimed at addressing the issue of so-called algorithmic discrimination, where systematic and repetitive errors in AI systems distort the processing of inputs and generate discriminatory outputs for workers.

The bill clearly references the legal principles established by the aforementioned judicial precedents.

Establishing an AI Observatory for Labor

As one of its innovations, the bill introduces a specific Observatory for AI Adoption in the Workplace, under the Ministry of Labor and Social Policies. This body will define strategies for using AI in the labor sector, monitor its impact, identify the most affected labor sectors, and promote AI-related training for both workers and employers.

The provision in paragraph 3 is, however, open to criticism, as it states that the objectives must not impose new or increased financial burdens on public finances, effectively shifting the costs of training onto private entities.

In summary, the government’s approach with this bill not only aims to establish the fundamental principles and objectives for AI usage in the workplace, but also introduces operational measures, such as the obligation for employers, contractors, and professionals to provide transparent information regarding the use of AI.

The establishment of an Observatory to promote the responsible use of automated systems in the labor sector is also commendable. The responsibilities and functions of this body will be determined by the Ministry of Labor once the bill is fully implemented.

If approved by Parliament, the AI Bill could come into effect and become directly enforceable in Italy’s legal system, potentially making Italy one of the first EU countries to adopt national legislation on artificial intelligence.