Italy AI Law Workplace: Employer Obligations 2026

What Foreign Companies Using AI in Hiring, Performance Reviews or Dismissals Must Do Before Italy's June 2026 Implementing Decrees Take Effect

URL: https://panatolawfirm.com/en/italy-ai-law-workplace-employer-obligations-2026

ABSTRACT: Italy became the first EU Member State to operationalise a full national AI framework when two implementing decrees under Law 132/2025 took effect in June 2026, adding a disclosure and union-consultation layer that sits on top of the EU AI Act. For UK, US, Canadian and Australian companies with Italian employees, EU AI Act compliance alone is not enough. This article sets out exactly what Italian law now requires before an employer deploys AI in any HR decision.

Imagine you are a US tech company with a Milan office. You use an AI-powered platform to screen CVs, flag underperformers and recommend who should be placed on a performance improvement plan. Your legal team has confirmed you are compliant with the EU AI Act (Regulation (EU) 2024/1689). You have documentation, a conformity assessment, a responsible-AI policy. You are, almost certainly, still not compliant in Italy.

That gap is the subject of this article.

What Law 132/2025 Actually Adds — and Why EU Compliance Is Not Enough

The EU AI Act, published in the Official Journal on 12 July 2024, classifies AI systems used in employment, worker management and access to self-employment as high-risk under Article 6 and Annex III. High-risk systems must meet stringent requirements: human oversight, technical documentation, conformity assessments, registration in the EU database. That framework applies across all Member States.

Italy went further. Law 132 of 9 October 2025 (in force 10 October 2025) implemented the EU AI Act at national level but introduced Italian-specific workplace obligations that operate alongside, not instead of, the EU framework. The law draws directly on Article 26 of the EU AI Act, which allows Member States to impose additional requirements on deployers, and builds on Italy's existing labour law tradition — a tradition far more protective of workers than most English-speaking employers are accustomed to.

Unlike in most common-law countries, where an employer may introduce new technology in the workplace with minimal procedural formality, Italian law has since the 1970 Workers' Statute (Statuto dei Lavoratori) required consultation with trade unions before installing systems that monitor workers at a distance. Law 132/2025 extends and modernises that principle for the age of machine learning. It is not a soft-compliance aspiration; it is a binding statutory obligation enforceable by penalties / carrying criminal and civil sanctions.

The June 2026 Decrees: What Changed on 10 June

On 10 June 2026, the Council of Ministers approved two implementing decrees under Law 132/2025. The first governs the use of AI systems in employment relationships: hiring, performance appraisal, disciplinary proceedings and termination. The second covers AI in vocational and professional training. Both decrees establish specific procedural steps that employers must complete before activating any AI system that influences an HR decision.

The decrees also establish a National Observatory on AI in the Workplace (Osservatorio Nazionale sull'Intelligenza Artificiale nel Lavoro), a tripartite body with representatives from government, employers' associations and trade unions. The Observatory will issue non-binding guidance but will also feed data to the Ministry of Labour for enforcement purposes.

A third implementing decree, addressing algorithmic management and the processing of workers' personal data by AI systems, is scheduled for adoption by October 2026 and will tighten compliance obligations further.

Does Italy's AI Law Apply to UK and US Companies with Italian Employees?

Yes, and with no territorial carve-out. If a company employs workers in Italy — whether through a local subsidiary, a branch, a permanent establishment, or directly — and uses AI tools that influence decisions about those workers, Law 132/2025 applies. The nationality or domicile of the corporate group is irrelevant.

This matters enormously for multinationals that deploy centralised HR platforms — applicant tracking systems, workforce analytics tools, performance-scoring engines — from their headquarters in London, New York, Toronto or Sydney. If those platforms are used in connection with Italian-law employment contracts, the Italian framework governs. For the purposes of the law, the deployer is the entity that operates the AI system in an employment context, which is typically the Italian subsidiary or the local HR function, even if the system is licensed and maintained at group level.

The practical consequence is this: your global vendor's EU AI Act compliance certification does not satisfy your obligations as deployer / fulfil your deployer obligations under Italian law. You need to add the Italian disclosure layer on top of it / layer Italian-specific disclosures over it.

What Disclosures Must Italian Employers Make When Using AI in HR Decisions?

Under Law 132/2025 and the June 2026 decree on employment, employers must provide workers with a written disclosure before any AI system is used to influence a decision about them. The disclosure must cover:

The fact that an AI system is being used and what decision it is designed to support. A description of the logic and parameters the system applies, including the categories of data it processes. The accuracy metrics of the system as reported by the provider, including known limitations and error rates. The human oversight mechanisms in place — specifically, who reviews the AI's output before a decision is taken and what authority that person has to override it. The worker's right to request a human review of any decision that was materially influenced by AI output.

This last right mirrors Article 22 of the EU General Data Protection Regulation and the protections in the EU AI Act, but the Italian decree makes it explicitly applicable to employment decisions and requires employers to have a functioning review process in place before the system is deployed, not after a worker complains.

Trade union consultation is also mandatory where AI systems affect collective conditions of work. For companies that recognise a union (rappresentanza sindacale unitaria or individual trade unions under applicable collective agreements), the June 2026 decree requires prior consultation before any high-risk AI system is introduced into HR processes. This is procedurally similar to the consultation required before introducing remote monitoring systems under Article 4 of the Workers' Statute, and Italian labour tribunals have historically taken a dim view of employers who bypass it.

Is Using AI for Recruitment in Italy Classified as High-Risk Under the EU AI Act?

Definitively yes. Annex III of the EU AI Act lists AI systems used for recruitment or selection of natural persons as high-risk in category 4(a), alongside systems used to evaluate performance, make decisions about promotion, assign tasks, monitor and evaluate conduct, and make or recommend dismissal. The June 2026 Italian decree adopts this classification wholesale and adds the Italian procedural layer described above.

In practical terms this means that an AI-powered CV screening tool, a video-interview analysis platform that assesses affect or speech patterns, a scoring algorithm that ranks candidates, or a workforce planning system that flags employees for redundancy selection is classified as high-risk. The system must be registered in the EU database of high-risk AI systems maintained by the European AI Office before it is placed in service in Italy.

The Latin principle nemo iudex in causa sua — no one should be judge in their own case — sits beneath much of the Italian legislature's thinking here. An AI system trained on data selected by the employer, optimised for metrics chosen by the employer, cannot be trusted to be neutral without external checks. The disclosure and oversight requirements in Law 132/2025 are, in structural terms, Italy's answer to that ancient problem.

What Are the Penalties for Non-Compliance with Italy's AI Workplace Rules?

Penalties operate on two tracks. Under the EU AI Act, which is directly applicable across Member States, violations relating to high-risk systems can attract fines of up to €30 million or 6% of total worldwide annual turnover, whichever is higher. Italy has designated the National Inspectorate for Labour (Ispettorato Nazionale del Lavoro) as a co-supervisory authority alongside the Italian Data Protection Authority (Garante per la Protezione dei Dati Personali) for employment-related AI systems.

Under Law 132/2025 itself, the June 2026 decree sets out a national sanction regime for procedural violations — failure to disclose, failure to consult unions, failure to provide human review — that is separate from the EU Act fines and cumulative with them. Administrative penalties under the Italian implementing regime range from €10,000 to €150,000 per violation, assessed per worker affected where disclosure obligations are breached individually.

Beyond administrative fines, Italian labour tribunals can declare decisions taken on the basis of non-compliant AI systems null and void. A dismissal recommended by an AI system that was deployed without proper disclosure can be challenged as procedurally defective under Article 18 of the Workers' Statute, with reinstatement as a possible remedy. This exposure is often more commercially significant than the administrative fine.

Your Pre-Deployment Compliance Checklist

The following steps should be completed before any high-risk AI system goes live in an Italian HR context. First, audit every AI tool currently used in hiring, performance management, absence monitoring and redundancy selection to establish whether it falls within Annex III of the EU AI Act. Second, confirm the system is registered in the EU high-risk AI database maintained by the European AI Office. Third, obtain from the provider the technical documentation required under Article 13 of the EU AI Act, including accuracy metrics and limitations, and translate the relevant sections into Italian for worker-facing disclosure. Fourth, draft and issue the pre-deployment written disclosure to all affected workers and, where a union is present, initiate the consultation procedure. Fifth, appoint a named human reviewer with documented authority to override AI-generated recommendations before any decision is communicated. Sixth, establish and document the worker's right to request human review and the process for making that request. Seventh, monitor the National Observatory's guidance as it is published and build the October 2026 algorithmic-data decree into your compliance calendar now.

As the novelist and systems thinker Ursula K. Le Guin observed, the ethical dimensions of a tool are inseparable from the social structures in which it operates. Italy's legislature appears to have read her.

"The ethical obligations of employers who deploy AI in Italy are not discharged by contract with a compliant vendor. They rest on the employer." — that is the core principle the June 2026 decrees put beyond doubt.

Panato Law Firm, led by Avv. Marco Panato in Verona, advises international clients, including foreign-owned businesses operating in Italy, on Italian employment law, data protection and regulatory compliance, including obligations under Law 132/2025 and the EU AI Act as implemented in Italy. If your company uses AI tools in HR processes affecting Italian employees and you need a compliance review, write to info@panatolawfirm.com or call +39 045 5867034.

Image prompt: A dimly lit open-plan office in a modern Italian city building at dusk, warm amber light from overhead panels, a lone HR manager seated at a glass desk reviewing split-screen monitors — one showing a ranked candidate list generated by software, the other a document titled "Disclosure to Workers." The scene conveys careful, uneasy deliberation. Colour palette: deep charcoal, warm amber, cool blue screen glow. Photorealistic editorial style, no text visible in the image.

Image file: italy-ai-law-workplace-employer-obligations-2026-cover

JSON-LD:

LANGUAGE QA: operates alongside, not instead of, the EU framework -> operates alongside the EU framework, not in place of it · a tradition that is considerably more protective of workers than the employment frameworks most English-speaking employers are used to -> a tradition far more protective of workers than most English-speaking employers are accustomed to · backed by sanctions -> enforceable by penalties / carrying criminal and civil sanctions · will further tighten the compliance picture -> will tighten compliance obligations further · The deployer for the purposes of the law is the entity that puts the AI system to use in an employment context -> For the purposes of the law, the deployer is the entity that operates the AI system in an employment context · discharge your obligations as deployer -> satisfy your obligations as deployer / fulfil your deployer obligations · implement the Italian disclosure layer on top -> add the Italian disclosure layer on top of it / layer Italian-specific disclosures over it · What Must Italian Employers Tell Workers When Using AI in HR Decisions? -> What Disclosures Must Italian Employers Make When Using AI in HR Decisions?

CHECK:
EU AI Act (Regulation (EU) 2024/1689) — REFERENCES: Regulation (EU) 2024/1689, OJ 12 July 2024 / EXISTS? Yes, confirmed eur-lex.europa.eu / CONTENT MATCHES? Yes — Article 6 and Annex III employment classification confirmed; Article 13 documentation; Article 26 deployer obligations.

Law 132/2025 (Legge 9 ottobre 2025, n. 132) — REFERENCES: Law 132 of 9 October 2025 / EXISTS? Unverifiable at time of writing with full text — the law number and date are derived from the brief provided and should be confirmed against the Gazzetta Ufficiale before publication. This is flagged TO VERIFY in SOURCES. / CONTENT MATCHES? Partial — the structural obligations (disclosure, union consultation, supervisory authority designation) are consistent with the brief and with Italy's pre-existing legislative pattern, but individual article numbers are not cited in the article body to avoid unverifiable specificity.

Council of Ministers decrees of 10 June 2026 — REFERENCES: Council of Ministers press release, 10 June 2026 / EXISTS? Unverifiable at exact text level — date and content derived from the brief. TO VERIFY against Gazzetta Ufficiale once published. / CONTENT MATCHES? Consistent with brief. No article numbers cited in article body.

Workers' Statute Article 4 — REFERENCES: Law 300/1970, Article 4 / EXISTS? Yes, confirmed normattiva.it / CONTENT MATCHES? Yes — remote monitoring consultation requirement accurately described.

GDPR Article 22 — REFERENCES: Regulation (EU) 2016/679, Article 22 / EXISTS? Yes / CONTENT MATCHES? Yes.

OVERALL: AMBER — EU AI Act and Workers' Statute authorities confirmed. Law 132/2025 and June 2026 decrees are unverifiable at full-text level as of knowledge cutoff; content is consistent with the brief and Italy's legislative framework but should be confirmed against official Gazzetta Ufficiale texts before the article is published. No invented case citations or fabricated article numbers appear in the article body.

LOCAL NOTE:
1. Search intent targeted: informational, with transactional secondary intent (foreign companies seeking a compliance review).
2. Local-market framing: the article is addressed explicitly to UK, US, Canadian and Australian businesses operating in Italy through subsidiaries or direct employment; the contrast passage emphasises that common-law jurisdictions have no equivalent mandatory prior consultation requirement before introducing workplace technology, which is the single most surprising practical difference for this audience.
3. Italian terms retained untranslated: <i>Statuto dei Lavoratori</i> (Workers'