Italy's GPDP has asked DeepSeek to explain the data it processes through its chatbot. The privacy regulator also wants to know what data DeepSeek's AI chatbot is trained with. The Italian Apple App Store and Google Play Store have removed the DeepSeek app, Italian news agency ANSA has reported.
The GPDP considers the DeepSeek app a potentially major risk to the data of millions of people in Italy. If personal data was collected with scraping, DeepSeek should inform users about how their data is processed.
Multiple U.S. governments have banned the use of the Chinese AI app because of potential security and ethical concerns. According to David Sacks, the current U.S. government's AI chief, there is evidence that DeepSeek has used OpenAI's AI models on a large scale to train its systems: "There is a technology called knowledge distillation, which allows one model to learn from the other model by asking millions of questions. This allows an AI to learn the other model's way of reasoning and basically suck a lot of knowledge from the other model."
The Autoriteit Persoonsgegevens AP) is also warning about DeepSeek. The Dutch privacy regulator is going to further investigate the transfer of personal data to China: "There must be sufficient guarantees that people keep a grip on their personal data and know what happens to it. The question is whether DeepSeek and other Chinese companies comply with this."
Click here for the message from the GPDP.
