Italy requires human decisions for AI-assisted dismissals

Since 23 October 2026, Italy has required employers using artificial intelligence to make a human decision when dismissing employees. Legislative Decree No. 179 enters into force on that date, following its publication on 8 October. Among other things, it adapts Italian law to the European regulation on AI, known as the AI Act.

The measure extends beyond dismissals. Article 41 of the decree covers final hiring decisions and contract changes, as well as performance evaluations and disciplinary sanctions.

AI can assist the employer, but the final decision rests with a natural person who has effective and autonomous authority. According to PwC Italy, this requires an independent assessment of the tool’s output, rather than a purely formal approval.

A dismissal made in violation of this rule is void, as are other decisions that breach it. The employee’s consent does not constitute an exception.

Before processing takes place, the employer must inform employees in accordance with the applicable obligations. An employee may request an intelligible explanation of a decision concerning them. It must be obtained with the involvement of a natural person and specify any influence of AI as well as the main parameters considered.

Preliminary screening of applications is subject to a different regime

The preliminary search for and selection of applications are excluded from the concept of a final decision under this rule, even when a candidate is eliminated from subsequent stages. The final hiring decision, however, must be entrusted to a person with effective and autonomous authority.

Other protections may nevertheless apply to an automated rejection. As the European Commission points out, the General Data Protection Regulation, or GDPR, regulates, under certain conditions, exclusively automated decisions that produce legal effects or similarly significant effects. It provides for exceptions and safeguards. A rejection of an application may fall under these protections depending on the circumstances.

AI risks are included in safety assessments

Article 42 requires AI systems to be taken into account in risk assessments when they influence work organisation, production rates or how work is carried out. It also covers decision-making processes relevant to safety.

The employer must inform and train employees on the specific risks and on prevention and protection measures.