After OpenAI made improvements to the data protection of its ChatGPT program, it is available again in Italy. Previously, the local regulator had blocked it with immediate effect to prevent the unwanted leakage of personal data.
Changes to ChatGPT
The changes to ChatGPT are noticeable and clearly aim to give users more autonomy over their own data. For example, it is now possible to switch to a private mode, in which search queries no longer show up in ChatGPT history and are not used for further training of the language model. Furthermore, an export function has been created with which users can get an overview of which personal data ChatGPT has stored and processed about them. It was also clarified that all histories are only stored for 30 days in order to be able to access them in case of suspected abuse. After 30 days, they will then be permanently deleted.
The Italian data protection authority has also called for stricter protection of minors. For this reason, users in Italy must henceforth click that they are at least 18 years old or at least 13 years old and have the permission of their legal guardians to use the service.
Business version planned
OpenAI has further announced a business version of ChatGPT. In this business version, users are to have even more control over input and data processing. In principle, it should be ensured here that inputs are not recorded in the history and not used for further training of the AI – which is the case in the normal version only in private mode. In addition, the business version should also be more strongly oriented to the needs of companies in other aspects. An incident at Samsung, for example, shows that this is important. At the beginning of April, employees of the company had a code for a previously unpublished project checked for errors via ChatGPT, which led to it being used for further training of the AI. However, this is not at all in the interest of Samsung, which wanted to keep the project secret during the development phase.