OpenAI has pledged to take measures to address the privacy concerns that led to the Italian data protection agency, Garante, temporarily banning the chatbot last week.
OpenAI took ChatGPT offline in Italy following allegations by Garante that it had breached privacy rules by failing to verify the age of its users and for the “massive collection and storage of personal data” without any legal basis. It also promised to be more transparent in the way it handles user data and verifies user age. It pledged to send Garante a document outlining the measures it plans to take to meet the agency’s requests.
Garante stated that it had no intention of putting a brake on AI development but insisted on respecting rules aimed at protecting the personal data of Italian and European citizens. It will evaluate the proposals made by OpenAI, although it is likely to take several days to assess the contents of the letter.
OpenAI, in a blog post explained that it was working to develop “nuanced policies against behavior that represents a genuine risk to people” and had taken steps to protect users’ privacy, including removing personal information from its datasets and fine-tuning models to reject user prompts asking for such information.
OpenAI did not respond to a request for comment on Garante’s statement. The company emphasized in its blog post that it does not use data for selling its services, advertising, or building profiles of people, but rather to make its models more helpful for users.
ChatGPT improves by further training on the conversations people have with it. Although some of its training data includes personal information available on the public internet, OpenAI is keen to ensure that its models learn about the world, not private individuals. The company also pledged to respond to individual requests to delete their data from its systems.
The sources for this piece include an article in Reuters.