The emergence of chatbots based on artificial intelligence has been followed by many controversies in recent months. Although many users believe that such tools can help increase productivity and save time; But experts express their concern about security and privacy issues. For example, ChatGPT conversation history and user data on this platform can include very sensitive items; As a result, the question will arise, what do the controllers of this service do with user data?
History of ChatGPT, a treasure trove of user information
Tools based on large linguistic models (LLM) usually use public but sometimes copyrighted data for their training. This data can also include conversations made by users with chatbots. For example, ChatGPT’s history contains a large volume of user chats that can be used for various purposes. This issue has caused the legislators to consider restrictions for companies providing such services. For example, after the OpenAI chatbot was banned in some parts of Europe, the company was forced to provide users with an option to stop data logging and delete ChatGPT history.
AI models typically use data such as commands entered by users, satisfaction levels with answers provided, and images to improve the accuracy of their tools. OpenAI claims that it never uses ChatGPT history or Dall-E’s text-to-image conversion service for advertising purposes and to display targeted and personalized headlines. It is also mentioned in the rules of these services that private information is deleted before saving the text of interactions. However, if you do not trust such statements, you can manually disable the recording of your data.
Many applications use the Application Programming Interface (API) provided by OpenAI to provide AI-based services. The company notes that it does not store data entered by third-party applications; Unless the user has activated the data saving option. Of course, even if the relevant option is enabled, the data will remain on the servers for only 30 days. Of course, in some cases, this process is violated; For example, by order of judicial authorities, information may be stored on the server for a longer period of time.
Who has access to OpenAI data?
OpenAI has officially announced that it will share the collected data with a set of trusted partners. These companies help OpenAI to provide its various services to users. Although no specific name is mentioned in the registered documents, one of them is undoubtedly Microsoft, which provides extensive financial and infrastructure support. The Redmonds are also the largest shareholders in this company and use its various technologies such as the GPT-4 model in their services, including the Bing chatbot.
It is worth noting that the necessary processes for conversations between users and chatbots are carried out on Microsoft Azure cloud servers, and according to the information posted on the OpenAI website, Microsoft can access the data stored on its servers. Also, a limited number of OpenAI employees have access to user information for data analysis; But no details about access levels and necessary prerequisites have been mentioned.
Is the data of OpenAI users well protected?
The answer to this question is not so easy and depends on how much users trust private companies like OpenAI and its partners, including Microsoft. Regardless of the main goals of these companies, it should be kept in mind that there is a history of information leakage in them; For example, in March, the ChatGPT chat history of some users was leaked. Also, in a similar case, the last four digits of some people’s credit cards were exposed by sending wrong emails.
OpenAI claims to have fixed all previous security problems and now in a special scheme it pays experts $20,000 to find each security problem. However, it is recommended that you refrain from sharing your personal information such as your full name, home or work address, contact number and bank details in conversation with AI chatbots. Also, if you have shared such data in the past, you can delete your user account in order to delete the registered information.
How to disable ChatGPT chat history?
As mentioned in the previous sections, users can disable the ChatGPT conversation history recording process. You can also deactivate the user account to delete all your information. To do this, just follow the steps below:
- Go to the ChatGPT website and log in to your account after typing your username and password.
- In the left corner of the screen, click on your profile picture and then select Settings.
- In the pop-up window that opens, click on the Show option next to the Data Controls heading.
- Now turn the switch next to Chat History & Training to disabled.
- If you want to backup ChatGPT chat history, you can click Export Data option. The file containing the text of the conversations will be sent to you in the form of an email.
- To delete all your chats in the OpenAI database, go back to the main menu and click on your profile picture again.
- Then select Clear conversations and confirm the process in the opened window.
- If you want to ensure the deletion of all your data, select the Delete account option from the Data Controls section.
- In the opened window, type the desired email and phrase and press the red key.
Note that once your OpenAI account is deleted, it will not be possible to restore it or access your ChatGPT chat history again. So it is recommended to get a backup before doing this. Also, in the future, you can create a new account again using the same email; Although the previous information inside it will not be accessible.
