They have accused the firm of breaking EU data protection law, and have given it 30 days to defend itself.

Italy's data watchdog said in a statement it had "told OpenAI off for breaking data protection law with ChatGPT's artificial intelligence platform". The watchdog banned the chatbot for a few weeks last year, becoming the first Western country to do so.

After that short ban, the watchdog said that the "evidence showed that OpenAI had broken the rules in the EU GDPR", the European Union's data protection law. It said:

"OpenAI can try to argue against the breaches within 30 days."

The watchdog said that when it makes its final decision, it will "look at what's going on" with a group of experts set up by the EU's main data regulator to help countries deal with ChatGPT's issues.

Last year, the Italian watchdog said OpenAI had no legal reason to collect and store loads of personal data to train the algorithms that make ChatGPT work.

It said it was not clear whose data was being collected.