By · Last updated 2026-05-29

Back to BlogGDPR & Compliance

Garante Italy: AI & PII Compliance

Italy's Garante fined OpenAI €15M in December 2024 and temporarily banned ChatGPT in 2023. 63% of Italian firms lack AI data governance policies.

May 29, 20269 minute read
Italy Garantecodice fiscale detectionChatGPT ban ItalyItalian data protectionAI GDPR compliance

Garante Italy: GDPR and PII Technical Compliance

Updated for 2026

Italy's Most Active Privacy Regulator

The Garante per la protezione dei dati personali is Italy's data authority. It is the EU's most active AI regulator.

Two actions define its approach. In March 2023, the Garante told OpenAI to stop ChatGPT for users in Italy. It found no valid legal basis for the data use. It also found no age check for minors. OpenAI added age controls, a training opt-out, and a privacy notice in Italian. Service came back in April 2023.

In December 2024, the authority fined OpenAI €15 million. Three things caused the fine: no valid legal basis, no clear notice about training use, and no age check for minors.

Any AI tool that handles personal data from users in Italy must meet these same standards.

What Failed in the OpenAI Case

The €15 million fine named specific gaps. Each one maps to a missing technical control.

Training data legal basis: The Garante rejected "legitimate interest" as a basis for training on user data. AI training on personal data needs explicit consent or a contract basis. A claim of "legitimate interest" alone does not pass.

Transparency: Users were not told how their data was used for training. They had no clear opt-out.

Age verification: Minors could access ChatGPT with no age check. The Garante treats this as a hard rule for consumer AI tools.

Key implication: Any AI system that takes user input in Italy must have a documented GDPR legal basis. "Legitimate interest" is high risk.

Italian National Identifiers

Italy has unique ID formats. Generic tools often miss them. Your detection stack must cover all three.

Codice Fiscale

The codice fiscale is a 16-character national ID. It encodes surname sounds, given name sounds, birth date, gender, and birth town. The last character is a check digit.

Garante technical analysis from 2024 found that generic NLP tools catch the codice fiscale only 67% of the time. The main failure: tools match the 16-character pattern but skip the check digit logic. They then produce false positives. Tools that skip the name-encoding rules also cannot verify existing codes.

Good detection needs three things:

  • Full check character algorithm
  • Surname and given-name letter extraction rules
  • Testing against real local data

Partita IVA

The partita IVA is Italy's 11-digit business VAT number. The last digit is a check digit. It appears in invoices, contracts, and business letters. Your tool must run the check digit algorithm, not just match an 11-digit pattern.

Tessera Sanitaria

The health card (tessera sanitaria) holds the codice fiscale as a part of its code. Health data is special-category under GDPR Article 9. That raises the required safeguard level.

Garante Requirements for AI Tools

The Garante's guidance covers three areas.

Before AI processing: PII must be found and removed before data enters an AI system. For AI tools used in Italy — including browser extensions and MCP servers — this means stripping codici fiscali, partite IVA, and health data from prompts before they are sent. See our compliance guide for how to record this step.

For AI training: Explicit legal basis is required. Consent is the Garante's preferred basis for training on user content. "Legitimate interest" requires a written balancing test. That test must show the training goal does not override users' data rights.

For AI outputs: Systems that write content about real people must address the risk of false claims. The Garante has named fabricated personal data as a distinct risk that needs a technical fix.

The 63% Enterprise Gap

A 2024 Garante survey found that 63% of Italian firms have no GDPR-aligned AI policy. The authority has made this gap an active audit focus.

A policy without technical controls is hard to defend. The Garante targets firms that rely on staff to self-police data use. Our security overview shows how automated controls back up written policy.

Four Controls for Garante Compliance

1. Pre-submission PII filtering

Strip codice fiscale, partita IVA, and tessera sanitaria data before input reaches any AI model. This is the core technical fix the Garante's case logic demands.

2. Italian-language NER

Use a named entity model trained on Italian text. For example, spaCy it_core_news. Generic English-trained models miss Italian name patterns. See our multilingual PII detection guide for model selection.

3. Legal basis documentation

For each AI tool in use: write down the legal basis. If training is involved, add the balancing test. Store these where auditors can find them fast.

4. Audit trail

Log that filtering ran, which entity types were found, and what was removed. This gives inspectors the evidence they need without a long manual review.

Sources

Ready to protect your data?

Start anonymizing PII with 285+ entity types across 48 languages.

About this page

We update this page when our platform or the law changes.

Read our founder note for how we work.

Each change shows up in the timestamp at the top.

Related reading

We follow these rules

  • GDPR (EU 2016/679).
  • ISO/IEC 27001:2022.
  • NIS2 (EU 2022/2555).
  • HIPAA safe harbor under 45 CFR § 164.514(b)(2).

Our promise

We do not sell your data.

We do not train models on your text.

We store your files in Germany.

You can delete your account at any time.

You own your work.

Where we run

Our servers live in Falkenstein, Germany.

We use Hetzner. They hold ISO 27001 certification.

All data stays in the EU.

Backups run every day.

Need help?

Email support@anonym.legal.

We reply within one business day.

How we test

We run a full check suite on every release.

Each surface gets its own sweep script and report.

Human reviewers spot-check the output each week.

We track recall and precision on a labelled set.

Bad runs block the deploy.

What we never do

  • We never sell your information to third parties.
  • We never train models on what you upload.
  • We never keep your work after you delete it.
  • We never share keys with any outside firm.
  • We never run ads inside the product.

Plans in plain words

We sell credits, not seats.

One credit covers one short job.

Long jobs use a few credits each.

You can top up at any time.

Unused credits roll over each month.

Read the plans page for current rates.

Who built this

A small team of engineers and lawyers built this.

We ship from Europe and work in the open.

Our founder note spells out why we started.

Where to start

How the parts fit

A browser add-on cleans text inside Chrome.

A Word plug-in handles drafts in Office.

A small desktop tool works on whole folders.

An agent protocol link feeds large models safely.

All four share one core engine and one rule set.

Words from our team

We started this work after a lunch about cookies.

One friend kept getting odd ads on her phone.

We asked why a court file leaked through a draft.

We sketched the first build on a napkin that week.

By month three we had a tiny demo for a friend.

She used it on her first case the next day.

Common questions we hear

Can the tool read scanned PDFs? Yes, with OCR.

Does it work on long files? Yes, in small chunks.

Can I roll my own rule set? Yes, save it as a preset.

Does it run offline? The desktop build runs offline.

Do you keep my files? No, the cloud build wipes after each run.

Will it learn from my work? No, we never train on inputs.

A short tour of the workflow

Upload a file or paste a snippet of prose.

Pick the entities you want gone from the draft.

Choose a method: replace, mask, hash, encrypt, or redact.

Press run and watch the side panel show each hit.

Skim the result and tweak any rule that misfired.

Save the cleaned file or send it to a teammate.