This is a first in the Western world. Italy on Friday (March 31st) blocked the ChatGPT artificial intelligence for fears related to the use of data, two months after banning another program marketed as a "virtual friend".

In a statement, the Italian Authority for the Protection of Personal Data warns that its decision has an "immediate effect" and accuses the chatbot of not respecting European regulations and not verifying the age of underage users.

>> ChatGPT: pausing AI "an existential issue"?

This decision will result in "the temporary limitation of the processing of Italian users' data vis-à-vis OpenAI", the company behind the application, according to the document.

ChatGPT appeared in November and was quickly taken over by users impressed by its ability to clearly answer difficult questions, write sonnets or computer code. Funded by the computer giant Microsoft, which has added it to several of its services, it is sometimes presented as a potential competitor to the Google search engine.

Data loss

The Italian institution points out that ChatGPT "suffered on March 20 a loss of data concerning users' conversations and payment information of subscribers to the paid service".

After the first reports, OpenAI briefly interrupted the service, then acknowledged a bug in a now-fixed third-party tool that affected 1.2% of its subscribers and an indefinite number of free users.

The authority also accuses it of "the absence of an information note to users, whose data is collected by OpenAI, but especially the absence of a legal basis justifying the collection and mass storage of personal data, in order to "train" the algorithms that operate the platform".

>> Read also: "ChatGPT artificial intelligence and the democratization of cybercrime"

In addition, while the robot is intended for people over the age of 13, it "emphasizes that the absence of any filter to verify the age of users exposes minors to answers that are absolutely non-compliant with their level of development."

The CNIL, the French gendarme of personal data, "has not received a complaint and has no similar procedure in progress," she said in a reaction to AFP.

However, she approached her Italian counterpart "in order to discuss the findings that have been made" and aims to "clarify the legal framework in the coming months".

The Replika application banned

The Italian institution had blocked in early February for similar reasons the application Replika, which offers to chat with a tailor-made avatar. Some users had complained of receiving messages and images that were too daring, bordering on sexual harassment.

Once again, the Autorité asks OpenAI to "communicate within 20 days the measures taken" to remedy this situation, "under penalty of a penalty of up to 20 million euros or up to 4% of annual worldwide turnover", the maximum provided for by the European Regulation on personal data (GDPR).

This case shows that the GDPR, which has already led to billions of dollars in fines for tech giants, could also become the enemy of new AIs that generate content.

According to Nello Cristianini, professor at the University of Bath (UK), "the most important considerations" are "the use without an appropriate legal basis of personal data for model training and the growing possibility of this data being inaccurately reproduced".

AI also has much deeper fears than the mere exploitation of personal data and the European Union is currently preparing a draft regulation that could be finalized by early 2024, for application a few years later.

Europol warned earlier this week that criminals were ready to take advantage of artificial intelligence to commit fraud and other cybercrimes.

ChatGPT was also blocked shortly after its release in several schools or universities around the world, after fears of cheating on exams, and companies advised their employees against using the app.

"We have seen employees provide the strategic plans of their company to ChatGPT to ask it to make a presentation by slides (presentation materials Editor's note). Idiots! Because all the data goes into ChatGPT which can regurgitate it if a competitor asks him the strategy of this company, "explained Françoise Soulie Fogelman, advisor of the FranceIA Hub, during a conference.

On Wednesday, billionaire Elon Musk – one of the founders of OpenAI from which he later left the board – and hundreds of global experts called for a six-month pause in research on AI more powerful than GPT-4, the latest version of the software on which ChatGPT is based launched in mid-March, citing "major risks for humanity".

With AFP

The summary of the week France 24 invites you to look back on the news that marked the week

I subscribe

Take international news with you everywhere! Download the France 24 app