A few days ago, the Italian Privacy Authority requested that OpenAI immediately stop managing personal data. This forced OpenAI to suspend ChatGPT in Italy.
I believe that the authority made several mistakes:
- There were no reasons for an urgent ban. The Authority did not provide a deadline to comply with local regulations. The ban was immediate and allowed only 20 days to demonstrate compliance in order to avoid a minimum fine of 20 million euros. - They did not consider the interests and needs of the citizens. - They confused the responsibility of the user using content generated by the tool with the tool itself. The Authority unreasonably requested that the tool never hallucinates when discussing people. They accused OpenAI of training on billions of people's personal data without notifying the data owners. In reality, they were referring to publicly available web data sources, such as CommonCrawl, Wikipedia, and others, which demonstrates a limited understanding of widely accepted standards for training data. For these reasons, I believe that the ChatGPT ban should be lifted immediately.