Italy unbans ChatGPT after OpenAI adds privacy controls

ChatGPT is officially back online in Italy after a month-long regulatory standoff.

Italian privacy watchdog Garante lifted its temporary ban on Friday after OpenAI rolled out a series of privacy controls and transparency updates. Italy shocked the tech world in late March by becoming the first Western nation to block the viral chatbot, accusing the company of training its models on personal data without proper legal grounds and failing to protect minors.

The concessions that got OpenAI unbanned

To get the green light, OpenAI had to meet several privacy conditions set by the Italian regulator.

First, users now have a way out. OpenAI added an opt-out toggle that lets people stop their chat histories from being used to train the company's underlying algorithms. The company also published clearer explanations on its site detailing how it processes and saves personal data.

Second, age verification is now in place for users accessing the site from Italy to screen out children under 13. Regulators had previously complained that young kids were exposed to completely unsuitable answers.

OpenAI also slapped a visible notice on the interface warning users that ChatGPT can generate inaccurate information about people, places, or facts. Interestingly, these privacy toggles have already surfaced outside Italy, including for users in Germany.

An OpenAI spokesperson said the company is delighted to welcome back Italian users and remains committed to protecting their personal data.

Why it matters

For everyday users, the biggest win here is control. You can finally tell ChatGPT to stop learning from your proprietary prompts and private questions with a simple toggle, a feature that should have been standard from day one.

We think OpenAI made a very smart tactical move by cooperating instantly instead of fighting a lengthy legal war. Europe’s regulatory framework is notoriously aggressive, and if Italy had maintained a total blockade, other European watchdogs would have likely followed suit within weeks, cutting OpenAI off from millions of active users.

That said, the fix still feels like a patch rather than a cure. Age screening via self-reported prompts or basic site checks is notoriously easy to bypass, and OpenAI still has no clear mechanism to purge existing personal data already baked into its existing neural weights.

This outcome provides the first real blueprint for how AI labs will have to operate in the Western world. Scrape first and apologize later will no longer fly, and transparent training opt-outs are about to become standard table stakes everywhere.