Italy fines OpenAI over ChatGPT privateness guidelines breach

By Elvira Pollina and Alvise Armellini

MILAN (Reuters) -Italy’s information safety company mentioned on Friday it fined ChatGPT maker OpenAI 15 million euros ($15.58 million) after closing an investigation into use of private information by the generative synthetic intelligence utility.

The high quality comes after the authority discovered OpenAI processed customers’ private information to “practice ChatGPT with out having an ample authorized foundation and violated the precept of transparency and the associated data obligations in the direction of customers”.

OpenAI mentioned the choice was “disproportionate” and that the corporate will file an attraction in opposition to it.

The investigation, which began in 2023, additionally concluded that the U.S.-based firm didn’t have an ample age verification system in place to stop youngsters underneath the age of 13 from being uncovered to inappropriate AI-generated content material, the authority mentioned.

The Italian watchdog additionally ordered OpenAI to launch a six-month marketing campaign on Italian media to boost public consciousness about how ChatGPT works, significantly near to information assortment of customers and non-users to coach algorithms.

Italy’s authority, often called Garante, is among the European Union’s most proactive regulators in assessing AI platform compliance with the bloc’s information privateness regime.

Final yr it briefly banned the usage of ChatGPT in Italy over alleged breaches of EU privateness guidelines.

The service was reactivated after Microsoft-backed OpenAI addressed points regarding, amongst different issues, the precise of customers to refuse consent for the usage of private information to coach the algorithms.

“They’ve since recognised our industry-leading strategy to defending privateness in AI, but this high quality is almost twenty occasions the income we made in Italy in the course of the related interval,” OpenAI mentioned, including the Garante’s strategy “undermines Italy’s AI ambitions”.

The regulator mentioned the dimensions of its 15-million-euro high quality was calculated taking into consideration OpenAI’s “cooperative stance”, suggesting the high quality may have been even larger.

Below the EU’s Common Knowledge Safety Regulation (GDPR) launched in 2018, any firm discovered to have damaged guidelines faces fines of as much as 20 million euros or 4% of its international turnover.

($1 = 0.9626 euros)