Your personal data could be exposed if you are not paying attention to this parameter.
Generative artificial intelligence, in particular Chatgpt, raises important questions in terms of data confidentiality. Recently, the organization of consumers and users (OCU), a consumer defense organization recognized in Spain, has issued a warning concerning a specific functionality of this practical tool. OCU members even recommend Chatgpt users to quickly deactivate a specific option.
The main concern of the OCU is the potential use of user conversation by OPENAI, the company developing Chatgpt. Because you should know that the platform is configured by default to be able to suck your conversations and your exchanges with the AI in order to train and improve its own models. And many users do not necessarily realize it.
The parameter that poses a problem is entitled “Improvement of the model”. It is available in the application and Chatgpt website settings. When activated, it allows not only the safeguarding of conversations in the personal user history, but also authorizes OpenAi to analyze your discussions to refine its algorithms. The OCS Alert on the use of sensitive personal information (health data, political opinions, financial information, etc.) by OPENAI. Because you may not want the company to know what your everyday habits and thoughts are.
It is however possible to modify this sharing sometimes a little too greedy. To do this, the user must go to the chatgpt settings, access the “Data Management” section, then deactivate the “Improve the model for all” option. Once this manipulation has been carried out, the new conversations will no longer be kept in the history and, crucially, will no longer be used by Openai. Note that the chatbot also offers an “ephemeral” mode, located at the top right of the window, which allows you not to use your discussions to train its AI model.
However, it should be noted that the deactivation of this option does not lead to the immediate deletion of conversations by Openai. In accordance with the company’s policy, these data is kept for a period of 30 days for the purposes of safety and detection of possible abuses, before their final deletion. Deactivation mainly prevents their use for continuous improvement in artificial intelligence.
In conclusion, the OCU recommends a verification of this parameter by Chatgpt users and its deactivation for those who wish to control their personal data. If necessary, all your information would end in the wild in the event of a safety breach or hack of the OpenAi servers.