AI and employment law in Italy: what does the new bill say and how should businesses implement it?

The Italian Council of Ministers, on 23 April, approved a bill to introduce provisions on AI, in line with the EU AI Act (European regulation on artificial intelligence approved on 13 March by the European Parliament). Two articles of the bill are dedicated to the use of artificial intelligence in labour matters (articles 10 and 11).

Labour provisions 

Article 10 clarifies that AI can be used to improve working conditions, protect the psycho-physical integrity of workers, increase the quality of work performance and the productivity of people, in accordance with EU law.

It specifies that the use of AI in the work environment must be safe, reliable, transparent and shall not violate human dignity or infringe the confidentiality of personal data. The employer or principal is obliged to inform the employee about the use of AI.

Finally, the principle of fairness in the management of the employment relationship is reaffirmed, stating that the use of AI systems may under no circumstances discriminate against workers.

Article 11 establishes, at the Ministry of Labour, an observatory on the adoption of artificial intelligence systems in the world of work, in order to maximise the benefits and contain the risks arising from the use of AI systems in the workplace, as well as to promote the training of workers and employers in AI. 
These provisions outline the regulatory criteria that should guide the use of AI in the world of work. But what exactly is being referred to? Focusing on the concrete application of AI in companies and the related challenges and ethical issues involved, this Insight also provides practical tips for interpreting the content of the bill and implementing it within companies.

AI applications in the world of work

In the world of work, AI can have multiple applications. These include recruiting, monitoring work performance, simplifying certain activities, and creating new roles. 

• Recruiting: AI can analyse the CVs of thousands of candidates in a few moments. Algorithms can identify the most suitable candidates based on keywords and specific skills required.
• Work performance monitoring: AI platforms can monitor employee performance, for example by analysing productivity data, and suggesting improvements. AI can also help companies monitoring compliance with labour regulations (such as in relation to hours worked, overtime, or holidays), as well as reporting violations of these regulations in real time.
• Simplification of certain tasks: AI can analyse huge amounts of data in the blink of an eye and make immediate decisions, absorbing the repetitive tasks performed by the worker by automating part of his work.
• Creation of new roles: Some examples of figures that will be increasingly in demand in the coming years are the data analyst, the cybersecurity analyst, the specialist in AI ethics or digital transformation, or the HR applied to AI.
   

The use of AI offers numerous advantages for employers: it improves organisational efficiency, reduces costs and minimises human error. It also provides valuable data for making informed decisions in the shortest possible time. 

Employees also benefit from AI, as they can receive immediate feedback on their applications, and simplify some of their work tasks, being relieved of repetitive, or even dangerous, tasks in order to concentrate on more value-added activities. 

Challenges and ethical issues

The use of AI, however, also presents challenges. These include: 

• IT security and privacy: For example, the unlawful processing of personal data and violation of the GDPR, or disclosure of confidential employee information, but also of business secrets and know-how.
• Control of workers: AI tools can continuously monitor workers' activities, which may lead to violations of Article 4 of Law 300/1970 on remote control.
• Unconscious biases: AI tools are not programmed to detect forms of discrimination or cognitive biases, but are based on historical data, thus risking to promote inequalities and indirect discrimination.
• Transparency of algorithms: decisions made by AI can be difficult for workers to understand (and possibly challenge).
• Accuracy of output: AI responses are not infallible, especially when it comes to law. Therefore, if we would entrust AI alone with the responsibility of producing output without human control, there would be a risk of inaccurate, fallacious or misleading results, with significant reputational damage for employers in the event of business crises.
• Dispersion of skills: entrusting only the AI with the responsibility of producing all kinds of output without human control could result in some human skills, no longer being exercised and trained, being lost over time. 

Company supervision?

As controversial as the use of AI may seem, it is no longer a question of accepting or opposing change. Change is already taking place and the best option is to learn about and manage AI in order to use it as an ally, in compliance with the AI Act.

In order to limit the risks, corporate supervision of the use of AI is essential. For instance: 

• implementing regulations for IT control and approval when acquiring new AI technologies;
• drafting policies to regulate the use of AI in the company, which can oversee the compliance of the use of such tools with current labour legislation; 
• verifying that the data used to "train" the algorithms are free of unconscious bias, including by implementing regular audits to identify and amend any discriminatory mechanisms;
• making sure that the algorithms are as transparent and explainable as possible, in case of possible objections from employees, and identifying clear ways to communicate the use of AI both inside and outside the company;
• identifying one or more people within the company to take responsibility for the outputs generated by AI;
• analysing, from a labour law point of view, the new roles connected with AI that will arise in companies, in order to identify their correct legal and contractual framework;
• providing training courses to be offered to employees to ensure the effective use of AI tools implemented in the company, as well as to accompany them in the acquisition of new skills and in the technology transition process. 
   

Osborne Clarke comment

The ultimate challenge of AI will not be the technology, but rather the ability of organisations to intercept, metabolise and wisely implement the potential of new technologies. 
In this, it will be crucial to create synergies between all actors involved: legislators, companies, workers and legal advisors.